{ "generated_at": "2026-10-17", "source": "research/models_2026.json + scripts/legacy-models.v1.json (legacy size facts)", "ranking_method": { "primary": "Descending live Artificial Analysis Intelligence Index, highest-scoring named evaluated mode per model (reasoning or non-reasoning). Provider-estimated scores stay visibly flagged.", "secondary": "Arena overall text ratings shown separately; tie-break only for a mapped non-proprietary entry. Broadly overlapping uncertainty means adjacent ranks are not proven differences.", "community": "Dated Reddit/HF evidence changes practical recommendations and caveats, not the numerical index. Community X fetches blocked; no X consensus invented.", "eligibility": "Verified public release within 2026 as of cutoff; exact day unknown if only month/year is sourced. Superseded same-product checkpoints retained without duplicate rank.", "scope": "Rank among researched eligible open-weight candidates, not a guarantee of exhaustive global best. Models on hosted benchmarks and local quantizations can differ.", "not_local_speed": "AA API token speeds/TTFT are not copied into the GPU/Mac speed database.", "no_regression": "No conversion or average between Arena Elo-like ratings and AA index." }, "research_counts": { "models": 100, "verified_2026": 99, "exact_release_dates": 82, "ranked": 75, "legacy": 47 }, "counts": { "models_total": 147, "current_2026": 97, "default_visible": 97, "ranked": 75, "exact_release_day": 82, "superseded": 2, "candidate": 1, "legacy": 47, "hardware": 163, "speeds": 4373, "speeds_site_matchable": 473 }, "models": [ { "slug": "mimo-v2.6-pro-mopd", "name": "MiMo-V2.6-Pro-MOPD", "hf_repo": "XiaomiMiMo/MiMo-V2.6-Pro-MOPD", "hf_revision": "6a7f202e8dbd0e57b3fe1da03917d08d75abc4b8", "source_url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-MOPD", "org": "XiaomiMiMo", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-09-27", "precision": "day", "url": "https://mimo.xiaomi.com/blog/mimo-v2-6-tool-call-repetition", "status": "verified_primary", "note": "MOPD checkpoint release; original series Sep21/22, API updated Sep25 06:00 UTC+8." }, "params": { "total_b": 1020, "active_b": 42, "moe": true, "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-MOPD/raw/6a7f202e8dbd0e57b3fe1da03917d08d75abc4b8/README.md", "active_url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-MOPD/raw/6a7f202e8dbd0e57b3fe1da03917d08d75abc4b8/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-MOPD/raw/6a7f202e8dbd0e57b3fe1da03917d08d75abc4b8/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-MOPD/blob/6a7f202e8dbd0e57b3fe1da03917d08d75abc4b8/LICENSE" }, "quality": { "rank": 1, "aa": { "value": 46.3242065310383, "variant": "MiMo-V2.6-Pro", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/mimo-v2-6-pro", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1490.6785014822, "lower": 1481.349339658, "upper": 1500.0076633063, "votes": 4075, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Same hosted MiMo family name after September checkpoint update; Arena exact evaluated checkpoint is unknown. Family-level comparator only, not a verified MOPD or local-quant score." } }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-MOPD/raw/6a7f202e8dbd0e57b3fe1da03917d08d75abc4b8/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": {}, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_Q8_0", "GGUF_FP16", "GGUF_BF16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [ "RL and MOPD are distinct checkpoints. API switched to MOPD on Sep25; hosted AA/Arena scores are not proven identical to this local checkpoint." ] }, { "slug": "glm-5.3", "name": "GLM-5.3", "hf_repo": "zai-org/GLM-5.3", "hf_revision": "ba32f8c19c8738e2c57c71798c601d477a7bf844", "source_url": "https://huggingface.co/zai-org/GLM-5.3", "org": "zai-org", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08-28", "precision": "day", "url": "https://x.com/Zai_org/status/2093354097122455713", "status": "publisher_statement_via_secondary_embed", "note": "Official statement embedded in https://gigazine.net/gsc_news/en/20260829-glm-5-3-open/; X unavailable directly. August29 JST corresponds to August28UTC." }, "params": { "total_b": null, "active_b": null, "moe": true, "url": null, "active_url": null, "scope": null }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/zai-org/GLM-5.3/raw/ba32f8c19c8738e2c57c71798c601d477a7bf844/config.json", "extended": null }, "license": { "value": "glm-5.3", "url": "https://huggingface.co/zai-org/GLM-5.3/blob/ba32f8c19c8738e2c57c71798c601d477a7bf844/LICENSE" }, "quality": { "rank": 2, "aa": { "value": 44.777392385614, "variant": "GLM-5.3 (Max)", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/glm-5-3", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1471.4560297865, "lower": 1466.2604238135, "upper": 1476.6516357595, "votes": 22543, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/zai-org/GLM-5.3/raw/ba32f8c19c8738e2c57c71798c601d477a7bf844/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "GGUF_Q8_0": { "bytes": 801357677216, "repo": "unsloth/GLM-5.3-GGUF", "revision": "346b3591c7f28d1a23716f97a065ecf12ec14771", "url": "https://huggingface.co/unsloth/GLM-5.3-GGUF/tree/346b3591c7f28d1a23716f97a065ecf12ec14771", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 17, "status": "measured_file_manifest" }, "GGUF_BF16": { "bytes": 1507988027936, "repo": "unsloth/GLM-5.3-GGUF", "revision": "346b3591c7f28d1a23716f97a065ecf12ec14771", "url": "https://huggingface.co/unsloth/GLM-5.3-GGUF/tree/346b3591c7f28d1a23716f97a065ecf12ec14771", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 33, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_FP16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "step-5-preview", "name": "Step 5 Preview", "hf_repo": "stepfun-ai/Step-5-Preview", "hf_revision": "364b3b03fed2477aa0953fe855a2c4c8c48ee3cc", "source_url": "https://huggingface.co/stepfun-ai/Step-5-Preview", "org": "stepfun-ai", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-10", "precision": "month", "url": "https://www.stepfun.com/step-5-preview", "status": "derived_from_publisher_schedule_and_observed_public_files", "note": "Publisher scheduled weights Oct15; files verified public onOct17. Exact actual public-access day not established; do not treat planned date as observed release." }, "params": { "total_b": 600, "active_b": 27, "moe": true, "url": "https://huggingface.co/stepfun-ai/Step-5-Preview/raw/364b3b03fed2477aa0953fe855a2c4c8c48ee3cc/README.md", "active_url": "https://huggingface.co/stepfun-ai/Step-5-Preview/raw/364b3b03fed2477aa0953fe855a2c4c8c48ee3cc/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/stepfun-ai/Step-5-Preview/raw/364b3b03fed2477aa0953fe855a2c4c8c48ee3cc/config.json", "extended": null }, "license": { "value": "other", "url": "https://huggingface.co/stepfun-ai/Step-5-Preview/blob/main/LICENSE.txt" }, "quality": { "rank": 3, "aa": { "value": 43.7343049141614, "variant": "Step 5 Preview", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/step-5", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1441.3906418931, "lower": 1433.9272272071, "upper": 1448.8540565792, "votes": 7116, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Hosted comparator only; Arena still labels this entry proprietary. Do not attribute its rating to the open checkpoint." } }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/stepfun-ai/Step-5-Preview/raw/364b3b03fed2477aa0953fe855a2c4c8c48ee3cc/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": {}, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_Q8_0", "GGUF_FP16", "GGUF_BF16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "kimi-k3", "name": "Kimi K3", "hf_repo": "moonshotai/Kimi-K3", "hf_revision": "f831ab66814297da540d832a5235f8e904f29d06", "source_url": "https://huggingface.co/moonshotai/Kimi-K3", "org": "moonshotai", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-07-27", "precision": "day", "url": "https://www.kimi.ai/help/agent/agent-overview", "status": "verified_primary", "note": "Official retrospective explicitly says full weights July27; July16 was product/API release." }, "params": { "total_b": 2800, "active_b": 104, "moe": true, "url": "https://huggingface.co/moonshotai/Kimi-K3/raw/f831ab66814297da540d832a5235f8e904f29d06/README.md", "active_url": "https://huggingface.co/moonshotai/Kimi-K3/raw/f831ab66814297da540d832a5235f8e904f29d06/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/moonshotai/Kimi-K3/raw/f831ab66814297da540d832a5235f8e904f29d06/config.json", "extended": null }, "license": { "value": "kimi-k3", "url": "https://huggingface.co/moonshotai/Kimi-K3/blob/f831ab66814297da540d832a5235f8e904f29d06/LICENSE" }, "quality": { "rank": 4, "aa": { "value": 43.5938229518782, "variant": "Kimi K3 (Max)", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/kimi-k3", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1474.6645918512, "lower": 1470.0866148603, "upper": 1479.242568842, "votes": 32795, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/moonshotai/Kimi-K3/raw/f831ab66814297da540d832a5235f8e904f29d06/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": {}, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_Q8_0", "GGUF_FP16", "GGUF_BF16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [ "Official checkpoint uses compressed-tensors MXFP4. HF tensor-element total is serialized/packed and must not replace official logical parameter count." ] }, { "slug": "glm-5.3-flash", "name": "GLM-5.3-Flash", "hf_repo": "zai-org/GLM-5.3-Flash", "hf_revision": "46a050bec13002ad29ee795c16e61cf4fa3c4a9c", "source_url": "https://huggingface.co/zai-org/GLM-5.3-Flash", "org": "zai-org", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08-26", "precision": "day", "url": "https://docs.z.ai/release-notes/new-released", "status": "verified_primary", "note": "Publisher model release; API and open checkpoint identity must stay explicit." }, "params": { "total_b": 320, "active_b": 18, "moe": true, "url": "https://huggingface.co/zai-org/GLM-5.3-Flash/raw/46a050bec13002ad29ee795c16e61cf4fa3c4a9c/README.md", "active_url": "https://huggingface.co/zai-org/GLM-5.3-Flash/raw/46a050bec13002ad29ee795c16e61cf4fa3c4a9c/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/zai-org/GLM-5.3-Flash/raw/46a050bec13002ad29ee795c16e61cf4fa3c4a9c/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/zai-org/GLM-5.3-Flash/blob/46a050bec13002ad29ee795c16e61cf4fa3c4a9c/LICENSE" }, "quality": { "rank": 5, "aa": { "value": 41.807466113455, "variant": "GLM-5.3-Flash", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/glm-5-3-flash", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1470.5296626496, "lower": 1465.6747641218, "upper": 1475.3845611774, "votes": 32925, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/zai-org/GLM-5.3-Flash/raw/46a050bec13002ad29ee795c16e61cf4fa3c4a9c/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "GGUF_Q8_0": { "bytes": 340981966208, "repo": "unsloth/GLM-5.3-Flash-GGUF", "revision": "a38483c8cd5df544f53d70fb281afe97369d5ab6", "url": "https://huggingface.co/unsloth/GLM-5.3-Flash-GGUF/tree/a38483c8cd5df544f53d70fb281afe97369d5ab6", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 8, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 1164010176, "url": "https://huggingface.co/unsloth/GLM-5.3-Flash-GGUF/resolve/a38483c8cd5df544f53d70fb281afe97369d5ab6/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 1128047296, "url": "https://huggingface.co/unsloth/GLM-5.3-Flash-GGUF/resolve/a38483c8cd5df544f53d70fb281afe97369d5ab6/mmproj-F16.gguf" } ] }, "GGUF_BF16": { "bytes": 641641064288, "repo": "unsloth/GLM-5.3-Flash-GGUF", "revision": "a38483c8cd5df544f53d70fb281afe97369d5ab6", "url": "https://huggingface.co/unsloth/GLM-5.3-Flash-GGUF/tree/a38483c8cd5df544f53d70fb281afe97369d5ab6", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 14, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 1164010176, "url": "https://huggingface.co/unsloth/GLM-5.3-Flash-GGUF/resolve/a38483c8cd5df544f53d70fb281afe97369d5ab6/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 1128047296, "url": "https://huggingface.co/unsloth/GLM-5.3-Flash-GGUF/resolve/a38483c8cd5df544f53d70fb281afe97369d5ab6/mmproj-F16.gguf" } ] }, "MLX_4bit": { "bytes": 203992076296, "repo": "mlx-community/GLM-5.3-Flash-4bit", "revision": "fa04cbc3622b16aa7475404060b05e506d81bd4e", "url": "https://huggingface.co/mlx-community/GLM-5.3-Flash-4bit/tree/fa04cbc3622b16aa7475404060b05e506d81bd4e", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 62, "status": "measured_file_manifest", "mixed": true }, "MLX_8bit": { "bytes": 334126804629, "repo": "pipenetwork/GLM-5.3-Flash-MLX-8bit", "revision": "dd6207803ecc59033bf9cb2aa81c1d1dc83676ef", "url": "https://huggingface.co/pipenetwork/GLM-5.3-Flash-MLX-8bit/tree/dd6207803ecc59033bf9cb2aa81c1d1dc83676ef", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 32, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_FP16" ], "community": [], "notes": [] }, { "slug": "ling-3.1-flash", "name": "Ling 3.1 Flash", "hf_repo": "inclusionAI/Ling-3.1-flash", "hf_revision": "a4ccd6b45c002ddb236f9cb77d52cc427482bf19", "source_url": "https://huggingface.co/inclusionAI/Ling-3.1-flash", "org": "inclusionAI", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026", "precision": "year", "url": "https://www.ant-ling.com/", "status": "verified_publisher_announcement_year", "note": "Publisher announcement September30; open weights appeared later in October. Exact open-weight day remains unknown; 2026 family provenance plus live MIT weights verified." }, "params": { "total_b": 560, "active_b": 25, "moe": true, "url": "https://huggingface.co/inclusionAI/Ling-3.1-flash/raw/a4ccd6b45c002ddb236f9cb77d52cc427482bf19/README.md", "active_url": "https://huggingface.co/inclusionAI/Ling-3.1-flash/raw/a4ccd6b45c002ddb236f9cb77d52cc427482bf19/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 262144, "url": "https://huggingface.co/inclusionAI/Ling-3.1-flash/raw/a4ccd6b45c002ddb236f9cb77d52cc427482bf19/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/inclusionAI/Ling-3.1-flash/blob/a4ccd6b45c002ddb236f9cb77d52cc427482bf19/LICENSE" }, "quality": { "rank": 6, "aa": { "value": 41.0906200397391, "variant": "Ling 3.1 Flash", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/ling-3-1-flash", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/inclusionAI/Ling-3.1-flash/raw/a4ccd6b45c002ddb236f9cb77d52cc427482bf19/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 357969521664, "repo": "bartowski/Ling-3.1-flash-GGUF", "revision": "de6623498835e4aefd8d60806da9d952078531b3", "url": "https://huggingface.co/bartowski/Ling-3.1-flash-GGUF/tree/de6623498835e4aefd8d60806da9d952078531b3", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 10, "status": "measured_file_manifest" }, "GGUF_Q5_K_M": { "bytes": 436319120672, "repo": "bartowski/Ling-3.1-flash-GGUF", "revision": "de6623498835e4aefd8d60806da9d952078531b3", "url": "https://huggingface.co/bartowski/Ling-3.1-flash-GGUF/tree/de6623498835e4aefd8d60806da9d952078531b3", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 12, "status": "measured_file_manifest" }, "GGUF_Q8_0": { "bytes": 608380104192, "repo": "bartowski/Ling-3.1-flash-GGUF", "revision": "de6623498835e4aefd8d60806da9d952078531b3", "url": "https://huggingface.co/bartowski/Ling-3.1-flash-GGUF/tree/de6623498835e4aefd8d60806da9d952078531b3", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 16, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_FP16", "GGUF_BF16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [ "Card advertises 1M context but config max_position_embeddings is 262144; do not silently substitute 1M in memory calculations." ] }, { "slug": "qwen3.8-2.4t-a95b", "name": "Qwen3.8-2.4T-A95B", "hf_repo": "Qwen/Qwen3.8-2.4T-A95B", "hf_revision": "207bd685a7e3696cfaff12ded7c6a7ea0f88c996", "source_url": "https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B", "org": "Qwen", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08-12", "precision": "day", "url": "https://github.com/QwenLM/Qwen3.8", "status": "verified_primary", "note": null }, "params": { "total_b": 2400, "active_b": 95, "moe": true, "url": "https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B/raw/207bd685a7e3696cfaff12ded7c6a7ea0f88c996/README.md", "active_url": "https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B/raw/207bd685a7e3696cfaff12ded7c6a7ea0f88c996/README.md", "scope": null }, "arch_label": "MoE · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B/raw/207bd685a7e3696cfaff12ded7c6a7ea0f88c996/config.json", "extended": null }, "license": { "value": "qwen3.8-max", "url": "https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B/blob/207bd685a7e3696cfaff12ded7c6a7ea0f88c996/LICENSE" }, "quality": { "rank": 7, "aa": { "value": 39.9461723105325, "variant": "Qwen3.8 2.4T A95B", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/qwen3-8-2-4t-a95b", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": 94208, "win_per_token": null, "win_tokens": null, "win_max": 0, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B/raw/207bd685a7e3696cfaff12ded7c6a7ea0f88c996/config.json", "dtype": "BF16", "note": "BF16/FP16 KV estimate for full-attention layers only, batch=1. Sliding-window terms and fixed recurrent state are separate; excludes alignment, graphs, buffers, vision and MTP/drafters." }, "artifacts": { "GGUF_Q8_0": { "bytes": 2600249563264, "repo": "unsloth/Qwen3.8-2.4T-A95B-GGUF", "revision": "567d3e6ac26c5474b18311e619c04350fb9a5556", "url": "https://huggingface.co/unsloth/Qwen3.8-2.4T-A95B-GGUF/tree/567d3e6ac26c5474b18311e619c04350fb9a5556", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 56, "status": "measured_file_manifest" }, "GGUF_BF16": { "bytes": 4893172677568, "repo": "unsloth/Qwen3.8-2.4T-A95B-GGUF", "revision": "567d3e6ac26c5474b18311e619c04350fb9a5556", "url": "https://huggingface.co/unsloth/Qwen3.8-2.4T-A95B-GGUF/tree/567d3e6ac26c5474b18311e619c04350fb9a5556", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 140, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_FP16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "qwen3.8-flash-next", "name": "Qwen3.8-Flash-Next", "hf_repo": "Qwen/Qwen3.8-Flash-Next", "hf_revision": "de4b8e4d43b917e7706784d8bb445c9af86a3540", "source_url": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next", "org": "Qwen", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08-26", "precision": "day", "url": "https://github.com/QwenLM/Qwen3.8-Flash-Next", "status": "verified_primary", "note": "Official News entry explicitly dates the release." }, "params": { "total_b": 180, "active_b": 6, "moe": true, "url": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next/raw/de4b8e4d43b917e7706784d8bb445c9af86a3540/README.md", "active_url": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next/raw/de4b8e4d43b917e7706784d8bb445c9af86a3540/README.md", "scope": null }, "arch_label": "MoE · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next/raw/de4b8e4d43b917e7706784d8bb445c9af86a3540/config.json", "extended": null }, "license": { "value": "qwen-community-1.0", "url": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next/blob/de4b8e4d43b917e7706784d8bb445c9af86a3540/LICENSE" }, "quality": { "rank": 8, "aa": { "value": 39.8223021864741, "variant": "Qwen3.8-Flash-Next", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/qwen3-8-flash-next", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next/raw/de4b8e4d43b917e7706784d8bb445c9af86a3540/config.json", "dtype": "BF16", "note": "12 QSA/full-attention blocks and36 linear-attention blocks; two KV heads with head_dim256, plus indexer KV head1/head_dim128/compression4/budget2048. QSA indexer and recurrent state require backend-specific accounting; generic GQA coefficient is not a verified total cache formula." }, "artifacts": { "GGUF_Q8_0": { "bytes": 188225033248, "repo": "unsloth/Qwen3.8-Flash-Next-GGUF", "revision": "766911a6b7369840a91dbcd95f9f997acaab6cd6", "url": "https://huggingface.co/unsloth/Qwen3.8-Flash-Next-GGUF/tree/766911a6b7369840a91dbcd95f9f997acaab6cd6", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 6, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 907542944, "url": "https://huggingface.co/unsloth/Qwen3.8-Flash-Next-GGUF/resolve/766911a6b7369840a91dbcd95f9f997acaab6cd6/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 904004000, "url": "https://huggingface.co/unsloth/Qwen3.8-Flash-Next-GGUF/resolve/766911a6b7369840a91dbcd95f9f997acaab6cd6/mmproj-F16.gguf" } ] }, "GGUF_BF16": { "bytes": 354029930496, "repo": "unsloth/Qwen3.8-Flash-Next-GGUF", "revision": "766911a6b7369840a91dbcd95f9f997acaab6cd6", "url": "https://huggingface.co/unsloth/Qwen3.8-Flash-Next-GGUF/tree/766911a6b7369840a91dbcd95f9f997acaab6cd6", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 8, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 907542944, "url": "https://huggingface.co/unsloth/Qwen3.8-Flash-Next-GGUF/resolve/766911a6b7369840a91dbcd95f9f997acaab6cd6/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 904004000, "url": "https://huggingface.co/unsloth/Qwen3.8-Flash-Next-GGUF/resolve/766911a6b7369840a91dbcd95f9f997acaab6cd6/mmproj-F16.gguf" } ] }, "MLX_4bit": { "bytes": 111519423247, "repo": "mlx-community/Qwen3.8-Flash-Next-4bit", "revision": "07b5dc6c54600a359b87f1e53e7adf6351c72a2c", "url": "https://huggingface.co/mlx-community/Qwen3.8-Flash-Next-4bit/tree/07b5dc6c54600a359b87f1e53e7adf6351c72a2c", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 22, "status": "measured_file_manifest", "mixed": true } }, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_FP16", "MLX_8bit" ], "community": [], "notes": [ "Qwen3.8-Flash hosted API is a production variant; open Flash-Next has 262144 native context. Do not equate the hosted 1M default with native config.", "125B backbone + 51B n-gram embedding + 4B MTP; 6B activated excludes embedding access; sum is storage scope." ] }, { "slug": "deepseek-v4.1-flash", "name": "DeepSeek V4.1 Flash", "hf_repo": "deepseek-ai/DeepSeek-V4.1-Flash", "hf_revision": "314113f8a7dadb388215396ba5fbee2c610a04d6", "source_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash", "org": "deepseek-ai", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-09-10", "precision": "day", "url": "https://api-docs.deepseek.com/updates/", "status": "verified_primary", "note": null }, "params": { "total_b": 552, "active_b": 16, "moe": true, "url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash/raw/314113f8a7dadb388215396ba5fbee2c610a04d6/README.md", "active_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash/raw/314113f8a7dadb388215396ba5fbee2c610a04d6/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash/raw/314113f8a7dadb388215396ba5fbee2c610a04d6/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash/blob/314113f8a7dadb388215396ba5fbee2c610a04d6/LICENSE" }, "quality": { "rank": 9, "aa": { "value": 39.456167472527, "variant": "DeepSeek V4.1 Flash (Max)", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/deepseek-v4-1-flash", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1462.8474205203, "lower": 1456.6471826179, "upper": 1469.0476584228, "votes": 12793, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash/raw/314113f8a7dadb388215396ba5fbee2c610a04d6/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": {}, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_Q8_0", "GGUF_FP16", "GGUF_BF16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [ "Causal encoder-decoder and engram embeddings; prefill/decode active parameter counts differ. Generic GQA or MLA cache formula is unsafe.", "552B backbone; activated 8B during prefill and 16B decode." ] }, { "slug": "muse-spark-open", "name": "Muse Spark Open (816B-A42B)", "hf_repo": "meta-models/Muse-Spark-Open-816B-A42B", "hf_revision": "0fb02478b02565b2ce7fc57e6051c91172774ac9", "source_url": "https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B", "org": "meta-models", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-10-15", "precision": "day", "url": "https://research.meta.ai/blog/muse-spark-open", "status": "verified_primary", "note": null }, "params": { "total_b": 816, "active_b": 42, "moe": true, "url": "https://research.meta.ai/blog/muse-spark-open", "active_url": "https://research.meta.ai/blog/muse-spark-open", "scope": null }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/mlx-community/Muse-Spark-Open-816B-A42B-4bit/raw/main/config.json", "extended": null }, "license": { "value": "muse-model-license", "url": "https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B/blob/0fb02478b02565b2ce7fc57e6051c91172774ac9/LICENSE.md" }, "quality": { "rank": 10, "aa": { "value": 38.8666662997805, "variant": "Muse Spark Open (Xhigh)", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/muse-spark-open", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/mlx-community/Muse-Spark-Open-816B-A42B-4bit/raw/main/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "GGUF_Q8_0": { "bytes": 871648188873, "repo": "ggml-org/Muse-Spark-Open-816B-A42B-GGUF", "revision": "ada96652107984f184027153a77ec250d67ed271", "url": "https://huggingface.co/ggml-org/Muse-Spark-Open-816B-A42B-GGUF/tree/ada96652107984f184027153a77ec250d67ed271", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 10, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-Muse-Spark-Open-816B-A42B-BF16.gguf", "bytes": 4097601888, "url": "https://huggingface.co/ggml-org/Muse-Spark-Open-816B-A42B-GGUF/resolve/ada96652107984f184027153a77ec250d67ed271/mmproj-Muse-Spark-Open-816B-A42B-BF16.gguf" }, { "path": "mmproj-Muse-Spark-Open-816B-A42B-F16.gguf", "bytes": 4095795552, "url": "https://huggingface.co/ggml-org/Muse-Spark-Open-816B-A42B-GGUF/resolve/ada96652107984f184027153a77ec250d67ed271/mmproj-Muse-Spark-Open-816B-A42B-F16.gguf" }, { "path": "mmproj-Muse-Spark-Open-816B-A42B-Q8_0.gguf", "bytes": 2183868768, "url": "https://huggingface.co/ggml-org/Muse-Spark-Open-816B-A42B-GGUF/resolve/ada96652107984f184027153a77ec250d67ed271/mmproj-Muse-Spark-Open-816B-A42B-Q8_0.gguf" } ] }, "GGUF_BF16": { "bytes": 1640264556374, "repo": "ggml-org/Muse-Spark-Open-816B-A42B-GGUF", "revision": "ada96652107984f184027153a77ec250d67ed271", "url": "https://huggingface.co/ggml-org/Muse-Spark-Open-816B-A42B-GGUF/tree/ada96652107984f184027153a77ec250d67ed271", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 22, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-Muse-Spark-Open-816B-A42B-BF16.gguf", "bytes": 4097601888, "url": "https://huggingface.co/ggml-org/Muse-Spark-Open-816B-A42B-GGUF/resolve/ada96652107984f184027153a77ec250d67ed271/mmproj-Muse-Spark-Open-816B-A42B-BF16.gguf" }, { "path": "mmproj-Muse-Spark-Open-816B-A42B-F16.gguf", "bytes": 4095795552, "url": "https://huggingface.co/ggml-org/Muse-Spark-Open-816B-A42B-GGUF/resolve/ada96652107984f184027153a77ec250d67ed271/mmproj-Muse-Spark-Open-816B-A42B-F16.gguf" }, { "path": "mmproj-Muse-Spark-Open-816B-A42B-Q8_0.gguf", "bytes": 2183868768, "url": "https://huggingface.co/ggml-org/Muse-Spark-Open-816B-A42B-GGUF/resolve/ada96652107984f184027153a77ec250d67ed271/mmproj-Muse-Spark-Open-816B-A42B-Q8_0.gguf" } ] }, "MLX_4bit": { "bytes": 532781033893, "repo": "mlx-community/Muse-Spark-Open-816B-A42B-4bit", "revision": "a87282dd8b816996c003dc934a5ee23591afd420", "url": "https://huggingface.co/mlx-community/Muse-Spark-Open-816B-A42B-4bit/tree/a87282dd8b816996c003dc934a5ee23591afd420", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 100, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_FP16", "MLX_8bit" ], "community": [], "notes": [ "Original card/config HTTP401 during direct fetch. Configuration reported from mlx-community conversion with provenance, not represented as original.", "Official primary launch blog; card/config gated." ] }, { "slug": "mimo-v2.6-flash-mopd", "name": "MiMo-V2.6-Flash-MOPD", "hf_repo": "XiaomiMiMo/MiMo-V2.6-Flash-MOPD", "hf_revision": "b12f5573350bb1704513fcf119fca48e498b0423", "source_url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-MOPD", "org": "XiaomiMiMo", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-09-27", "precision": "day", "url": "https://mimo.xiaomi.com/blog/mimo-v2-6-tool-call-repetition", "status": "verified_primary", "note": "MOPD checkpoint release; original series Sep21/22, API updated Sep25 06:00 UTC+8." }, "params": { "total_b": 309, "active_b": 15, "moe": true, "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-MOPD/raw/b12f5573350bb1704513fcf119fca48e498b0423/README.md", "active_url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-MOPD/raw/b12f5573350bb1704513fcf119fca48e498b0423/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-MOPD/raw/b12f5573350bb1704513fcf119fca48e498b0423/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-MOPD/blob/b12f5573350bb1704513fcf119fca48e498b0423/LICENSE" }, "quality": { "rank": 11, "aa": { "value": 37.8843590141754, "variant": "MiMo-V2.6-Flash", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/mimo-v2-6-flash", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1455.2141256063, "lower": 1448.7630895366, "upper": 1461.665161676, "votes": 10917, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Same hosted MiMo family name after September checkpoint update; Arena exact evaluated checkpoint is unknown. Family-level comparator only, not a verified MOPD or local-quant score." } }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-MOPD/raw/b12f5573350bb1704513fcf119fca48e498b0423/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": {}, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_Q8_0", "GGUF_FP16", "GGUF_BF16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [ "RL and MOPD are distinct checkpoints. API switched to MOPD on Sep25; hosted AA/Arena scores are not proven identical to this local checkpoint." ] }, { "slug": "deepseek-v4-pro-0813", "name": "DeepSeek V4 Pro 0813", "hf_repo": "deepseek-ai/DeepSeek-V4-Pro-0813", "hf_revision": "72e1d3230f6c080a530b0a1d46f8eb4602340597", "source_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro-0813", "org": "deepseek-ai", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08-13", "precision": "day", "url": "https://api-docs.deepseek.com/updates/", "status": "verified_primary", "note": null }, "params": { "total_b": null, "active_b": null, "moe": true, "url": null, "active_url": null, "scope": null }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro-0813/raw/72e1d3230f6c080a530b0a1d46f8eb4602340597/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro-0813/blob/72e1d3230f6c080a530b0a1d46f8eb4602340597/LICENSE" }, "quality": { "rank": 12, "aa": { "value": 35.9967791278402, "variant": "DeepSeek V4 Pro 0813 (Max)", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/deepseek-v4-pro", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1445.5501558251, "lower": 1438.9816190908, "upper": 1452.1186925594, "votes": 9800, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro-0813/raw/72e1d3230f6c080a530b0a1d46f8eb4602340597/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": {}, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_Q8_0", "GGUF_FP16", "GGUF_BF16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "qwen3.8-27b", "name": "Qwen3.8-27B", "hf_repo": "Qwen/Qwen3.8-27B", "hf_revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0", "source_url": "https://huggingface.co/Qwen/Qwen3.8-27B", "org": "Qwen", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08-14", "precision": "day", "url": "https://github.com/QwenLM/Qwen3.8/blob/main/README.md", "status": "verified_primary", "note": "Official dated release-news entry states public model availability." }, "params": { "total_b": 27, "active_b": 27, "moe": false, "url": "https://huggingface.co/Qwen/Qwen3.8-27B/blob/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0/README.md", "active_url": null, "scope": "Publisher nominal language-model size; HF total includes stored multimodal/MTP tensors as applicable." }, "arch_label": "Dense · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/Qwen/Qwen3.8-27B/blob/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0/config.json", "extended": 1000000 }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/Qwen/Qwen3.8-27B/blob/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0/LICENSE" }, "quality": { "rank": 13, "aa": { "value": 33.6762858057476, "variant": "Qwen3.8 27B (Xhigh)", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/qwen3-8-27b", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1440.5473201399, "lower": 1435.4794995094, "upper": 1445.6151407704, "votes": 23294, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": 65536, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3.8-27B/blob/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 17442399968, "repo": "bartowski/Qwen3.8-27B-GGUF", "revision": "87410e7b9a5c21e6ca70c0c9cc5237de909fe621", "url": "https://huggingface.co/bartowski/Qwen3.8-27B-GGUF/tree/87410e7b9a5c21e6ca70c0c9cc5237de909fe621", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-Qwen3.8-27B-bf16.gguf", "bytes": 931145952, "url": "https://huggingface.co/bartowski/Qwen3.8-27B-GGUF/blob/87410e7b9a5c21e6ca70c0c9cc5237de909fe621/mmproj-Qwen3.8-27B-bf16.gguf" }, { "path": "mmproj-Qwen3.8-27B-f16.gguf", "bytes": 927607008, "url": "https://huggingface.co/bartowski/Qwen3.8-27B-GGUF/blob/87410e7b9a5c21e6ca70c0c9cc5237de909fe621/mmproj-Qwen3.8-27B-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 20923877088, "repo": "bartowski/Qwen3.8-27B-GGUF", "revision": "87410e7b9a5c21e6ca70c0c9cc5237de909fe621", "url": "https://huggingface.co/bartowski/Qwen3.8-27B-GGUF/tree/87410e7b9a5c21e6ca70c0c9cc5237de909fe621", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-Qwen3.8-27B-bf16.gguf", "bytes": 931145952, "url": "https://huggingface.co/bartowski/Qwen3.8-27B-GGUF/blob/87410e7b9a5c21e6ca70c0c9cc5237de909fe621/mmproj-Qwen3.8-27B-bf16.gguf" }, { "path": "mmproj-Qwen3.8-27B-f16.gguf", "bytes": 927607008, "url": "https://huggingface.co/bartowski/Qwen3.8-27B-GGUF/blob/87410e7b9a5c21e6ca70c0c9cc5237de909fe621/mmproj-Qwen3.8-27B-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 29116388960, "repo": "bartowski/Qwen3.8-27B-GGUF", "revision": "87410e7b9a5c21e6ca70c0c9cc5237de909fe621", "url": "https://huggingface.co/bartowski/Qwen3.8-27B-GGUF/tree/87410e7b9a5c21e6ca70c0c9cc5237de909fe621", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-Qwen3.8-27B-bf16.gguf", "bytes": 931145952, "url": "https://huggingface.co/bartowski/Qwen3.8-27B-GGUF/blob/87410e7b9a5c21e6ca70c0c9cc5237de909fe621/mmproj-Qwen3.8-27B-bf16.gguf" }, { "path": "mmproj-Qwen3.8-27B-f16.gguf", "bytes": 927607008, "url": "https://huggingface.co/bartowski/Qwen3.8-27B-GGUF/blob/87410e7b9a5c21e6ca70c0c9cc5237de909fe621/mmproj-Qwen3.8-27B-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 16054541349, "repo": "mlx-community/Qwen3.8-27B-4bit", "revision": "10c35caafbb80f7dc6a7a432cdd11af10a6d4818", "url": "https://huggingface.co/mlx-community/Qwen3.8-27B-4bit/tree/10c35caafbb80f7dc6a7a432cdd11af10a6d4818", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 3, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 29501218479, "repo": "mlx-community/Qwen3.8-27B-8bit", "revision": "815b83c0df8ffd1d1b5244cf75fd6ef14fca9ef9", "url": "https://huggingface.co/mlx-community/Qwen3.8-27B-8bit/tree/815b83c0df8ffd1d1b5244cf75fd6ef14fca9ef9", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 6, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-10-16", "title": "Qwen3.8-27B-Flash-MTP-GGUF--5.79 GB. Code ~70-90 t/s, editing ~100-200 t/s.", "url": "https://huggingface.co/Qwen/Qwen3.8-27B/discussions/213" }, { "platform": "Hugging Face", "date": "2026-10-12", "title": "Intermittent Persian language-instruction failures after tool results — Alibaba via OpenRouter", "url": "https://huggingface.co/Qwen/Qwen3.8-27B/discussions/212" }, { "platform": "Reddit", "date": "2026-09-15", "title": null, "url": "https://www.reddit.com/r/LocalLLaMA/comments/1wha6sv/i_benchmarked_ifmk2horizon7b_on_16gb_vram/" }, { "platform": "Reddit", "date": "2026-10-11", "title": null, "url": "https://www.reddit.com/r/LocalLLaMA/comments/1x37q7b/update_qwen_38_27b_140_toks_on_single_rtx_3090/" }, { "platform": "Reddit", "date": "2026-10-15", "title": null, "url": "https://www.reddit.com/r/LocalLLM/comments/1x70juc/qwen38_27b_took_7_hours_on_a_task_and_didnt_make/" } ], "notes": [] }, { "slug": "motif-3", "name": "Motif 3", "hf_repo": "Motif-Technologies/Motif-3", "hf_revision": "2ed2ed5cfabffa10fdabb2fc0d0288f8e6de893a", "source_url": "https://huggingface.co/Motif-Technologies/Motif-3", "org": "Motif-Technologies", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08", "precision": "month", "url": "https://arxiv.org/abs/2608.09119", "status": "verified_publisher_release_month", "note": "Publisher finished-model launch August13; exact open-weight day not independently pinned. Paper v1 August10, HF weights observed October17." }, "params": { "total_b": 314, "active_b": 13.2, "moe": true, "url": "https://huggingface.co/Motif-Technologies/Motif-3/raw/2ed2ed5cfabffa10fdabb2fc0d0288f8e6de893a/README.md", "active_url": "https://huggingface.co/Motif-Technologies/Motif-3/raw/2ed2ed5cfabffa10fdabb2fc0d0288f8e6de893a/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 262144, "url": "https://huggingface.co/Motif-Technologies/Motif-3/raw/2ed2ed5cfabffa10fdabb2fc0d0288f8e6de893a/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/Motif-Technologies/Motif-3/blob/2ed2ed5cfabffa10fdabb2fc0d0288f8e6de893a/LICENSE" }, "quality": { "rank": 14, "aa": { "value": 33.566562209723, "variant": "Motif 3", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/motif-3", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/Motif-Technologies/Motif-3/raw/2ed2ed5cfabffa10fdabb2fc0d0288f8e6de893a/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 197005412640, "repo": "AesSedai/Motif-3-GGUF", "revision": "653a2aff374d51c8e62085d20e40b3f057355c9e", "url": "https://huggingface.co/AesSedai/Motif-3-GGUF/tree/653a2aff374d51c8e62085d20e40b3f057355c9e", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 5, "status": "measured_file_manifest" }, "GGUF_Q5_K_M": { "bytes": 236311284192, "repo": "AesSedai/Motif-3-GGUF", "revision": "653a2aff374d51c8e62085d20e40b3f057355c9e", "url": "https://huggingface.co/AesSedai/Motif-3-GGUF/tree/653a2aff374d51c8e62085d20e40b3f057355c9e", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 6, "status": "measured_file_manifest" }, "MLX_8bit": { "bytes": 334649575273, "repo": "avlp12/Motif-3-Alis-MLX-8bit", "revision": "cfabac275cc14b6cf0872fa196009fd5be28b8a5", "url": "https://huggingface.co/avlp12/Motif-3-Alis-MLX-8bit/tree/cfabac275cc14b6cf0872fa196009fd5be28b8a5", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 78, "status": "measured_file_manifest", "mixed": true } }, "artifacts_missing": [ "GGUF_Q8_0", "GGUF_FP16", "GGUF_BF16", "MLX_4bit" ], "community": [], "notes": [] }, { "slug": "k2-horizon-375b-a23b", "name": "K2-Horizon-375B-A23B", "hf_repo": "IFM/K2-Horizon-375B-A23B", "hf_revision": "78272deca7adc2f90d154288b5161172283c80b3", "source_url": "https://huggingface.co/IFM/K2-Horizon-375B-A23B", "org": "IFM", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-09-03", "precision": "day", "url": "https://ifm.ai/blog/k2/", "status": "verified_primary", "note": null }, "params": { "total_b": 375, "active_b": 23, "moe": true, "url": "https://huggingface.co/IFM/K2-Horizon-375B-A23B/raw/78272deca7adc2f90d154288b5161172283c80b3/README.md", "active_url": "https://huggingface.co/IFM/K2-Horizon-375B-A23B/raw/78272deca7adc2f90d154288b5161172283c80b3/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 524288, "url": "https://huggingface.co/IFM/K2-Horizon-375B-A23B/raw/78272deca7adc2f90d154288b5161172283c80b3/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/IFM/K2-Horizon-375B-A23B/blob/78272deca7adc2f90d154288b5161172283c80b3/LICENSE" }, "quality": { "rank": 15, "aa": { "value": 30.5015302696723, "variant": "K2 Horizon 375B A23B", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/k2-horizon-375b-a23b", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": 249856, "win_per_token": null, "win_tokens": null, "win_max": 0, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/IFM/K2-Horizon-375B-A23B/raw/78272deca7adc2f90d154288b5161172283c80b3/config.json", "dtype": "BF16", "note": "BF16/FP16 KV estimate for full-attention layers only, batch=1. Sliding-window terms and fixed recurrent state are separate; excludes alignment, graphs, buffers, vision and MTP/drafters." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 237797004832, "repo": "bartowski/K2-Horizon-375B-A23B-GGUF", "revision": "be1b1021b0ec2d701bef519ea3a3ded77610678f", "url": "https://huggingface.co/bartowski/K2-Horizon-375B-A23B-GGUF/tree/be1b1021b0ec2d701bef519ea3a3ded77610678f", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 7, "status": "measured_file_manifest" }, "GGUF_Q5_K_M": { "bytes": 289588270656, "repo": "bartowski/K2-Horizon-375B-A23B-GGUF", "revision": "be1b1021b0ec2d701bef519ea3a3ded77610678f", "url": "https://huggingface.co/bartowski/K2-Horizon-375B-A23B-GGUF/tree/be1b1021b0ec2d701bef519ea3a3ded77610678f", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 8, "status": "measured_file_manifest" }, "GGUF_Q8_0": { "bytes": 403079845472, "repo": "bartowski/K2-Horizon-375B-A23B-GGUF", "revision": "be1b1021b0ec2d701bef519ea3a3ded77610678f", "url": "https://huggingface.co/bartowski/K2-Horizon-375B-A23B-GGUF/tree/be1b1021b0ec2d701bef519ea3a3ded77610678f", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 11, "status": "measured_file_manifest" }, "GGUF_BF16": { "bytes": 758484189216, "repo": "Baekpica/K2-Horizon-375B-A23B-GGUF", "revision": "81a5c2a52adfc817106c2a880e836b52bd92b96f", "url": "https://huggingface.co/Baekpica/K2-Horizon-375B-A23B-GGUF/tree/81a5c2a52adfc817106c2a880e836b52bd92b96f", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 30, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_FP16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "reflection-beam", "name": "Reflection Beam 1.0 501B-A23B", "hf_repo": "reflection/Beam-1.0-501B-A23B", "hf_revision": "78f8f654d40a40919b8e9954b40dd879ecb9a692", "source_url": "https://huggingface.co/reflection/Beam-1.0-501B-A23B", "org": "reflection", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-10-16", "precision": "day", "url": "https://reflection.ai/blog/beam-is-available", "status": "verified_primary", "note": "Weights released Oct16; Oct5 was announcement." }, "params": { "total_b": 501.3, "active_b": 23.4, "moe": true, "url": "https://huggingface.co/reflection/Beam-1.0-501B-A23B/raw/78f8f654d40a40919b8e9954b40dd879ecb9a692/README.md", "active_url": "https://huggingface.co/reflection/Beam-1.0-501B-A23B/raw/78f8f654d40a40919b8e9954b40dd879ecb9a692/README.md", "scope": null }, "arch_label": "MoE · sliding-window attn", "context": { "native": 1048576, "url": "https://huggingface.co/reflection/Beam-1.0-501B-A23B/raw/78f8f654d40a40919b8e9954b40dd879ecb9a692/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/reflection/Beam-1.0-501B-A23B/blob/78f8f654d40a40919b8e9954b40dd879ecb9a692/LICENSE" }, "quality": { "rank": 16, "aa": { "value": 30.3334426348039, "variant": "Beam 1.0", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/reflection-beam", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": 36864, "win_per_token": null, "win_tokens": null, "win_max": 90177536, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/reflection/Beam-1.0-501B-A23B/raw/78f8f654d40a40919b8e9954b40dd879ecb9a692/config.json", "dtype": "BF16", "note": "BF16/FP16 KV estimate for full-attention layers only, batch=1. Sliding-window terms and fixed recurrent state are separate; excludes alignment, graphs, buffers, vision and MTP/drafters." }, "artifacts": { "GGUF_Q8_0": { "bytes": 532794735998, "repo": "unsloth/Beam-1.0-501B-A23B-GGUF", "revision": "4cc80272381734e2b16b34c1af3362c7268a0ebc", "url": "https://huggingface.co/unsloth/Beam-1.0-501B-A23B-GGUF/tree/4cc80272381734e2b16b34c1af3362c7268a0ebc", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 13, "status": "measured_file_manifest" }, "GGUF_BF16": { "bytes": 1002671882528, "repo": "unsloth/Beam-1.0-501B-A23B-GGUF", "revision": "4cc80272381734e2b16b34c1af3362c7268a0ebc", "url": "https://huggingface.co/unsloth/Beam-1.0-501B-A23B-GGUF/tree/4cc80272381734e2b16b34c1af3362c7268a0ebc", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 23, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_FP16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [ "Publisher architecture table gives501.3B/23.4B; model name rounds to501B/23B." ] }, { "slug": "minimax-m3", "name": "MiniMax-M3", "hf_repo": "MiniMaxAI/MiniMax-M3", "hf_revision": "f0e1c1e04d40177e4673a22097036854f536e9c0", "source_url": "https://huggingface.co/MiniMaxAI/MiniMax-M3", "org": "MiniMaxAI", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-06", "precision": "month", "url": "https://arxiv.org/abs/2606.13392", "status": "verified_publisher_release_month", "note": "Publisher paper v1 June11 states model publicly released; community records actual weight release June12. API announced June1. Keep exact public-weight day unknown." }, "params": { "total_b": 428, "active_b": 23, "moe": true, "url": "https://huggingface.co/MiniMaxAI/MiniMax-M3/raw/f0e1c1e04d40177e4673a22097036854f536e9c0/README.md", "active_url": "https://huggingface.co/MiniMaxAI/MiniMax-M3/raw/f0e1c1e04d40177e4673a22097036854f536e9c0/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/MiniMaxAI/MiniMax-M3/raw/f0e1c1e04d40177e4673a22097036854f536e9c0/config.json", "extended": null }, "license": { "value": "minimax-community", "url": "https://huggingface.co/MiniMaxAI/MiniMax-M3/blob/f0e1c1e04d40177e4673a22097036854f536e9c0/LICENSE" }, "quality": { "rank": 17, "aa": { "value": 29.2202932236316, "variant": "MiniMax-M3", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/minimax-m3", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1433.3266154552, "lower": 1429.5115902011, "upper": 1437.1416407094, "votes": 63973, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/MiniMaxAI/MiniMax-M3/raw/f0e1c1e04d40177e4673a22097036854f536e9c0/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 261282110272, "repo": "bartowski/MiniMax-M3-GGUF", "revision": "2ee48e5fcf21a2522ee2fb9ffaa15592ec5498e6", "url": "https://huggingface.co/bartowski/MiniMax-M3-GGUF/tree/2ee48e5fcf21a2522ee2fb9ffaa15592ec5498e6", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 7, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-MiniMax-M3-bf16.gguf", "bytes": 1735294560, "url": "https://huggingface.co/bartowski/MiniMax-M3-GGUF/resolve/2ee48e5fcf21a2522ee2fb9ffaa15592ec5498e6/mmproj-MiniMax-M3-bf16.gguf" }, { "path": "mmproj-MiniMax-M3-f16.gguf", "bytes": 1732284000, "url": "https://huggingface.co/bartowski/MiniMax-M3-GGUF/resolve/2ee48e5fcf21a2522ee2fb9ffaa15592ec5498e6/mmproj-MiniMax-M3-f16.gguf" } ] }, "GGUF_Q5_K_M": { "bytes": 305348746176, "repo": "bartowski/MiniMax-M3-GGUF", "revision": "2ee48e5fcf21a2522ee2fb9ffaa15592ec5498e6", "url": "https://huggingface.co/bartowski/MiniMax-M3-GGUF/tree/2ee48e5fcf21a2522ee2fb9ffaa15592ec5498e6", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 8, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-MiniMax-M3-bf16.gguf", "bytes": 1735294560, "url": "https://huggingface.co/bartowski/MiniMax-M3-GGUF/resolve/2ee48e5fcf21a2522ee2fb9ffaa15592ec5498e6/mmproj-MiniMax-M3-bf16.gguf" }, { "path": "mmproj-MiniMax-M3-f16.gguf", "bytes": 1732284000, "url": "https://huggingface.co/bartowski/MiniMax-M3-GGUF/resolve/2ee48e5fcf21a2522ee2fb9ffaa15592ec5498e6/mmproj-MiniMax-M3-f16.gguf" } ] }, "GGUF_Q8_0": { "bytes": 453611165056, "repo": "unsloth/MiniMax-M3-GGUF", "revision": "abdce34b5ab2e7dd9260130415c01a1f25561210", "url": "https://huggingface.co/unsloth/MiniMax-M3-GGUF/tree/abdce34b5ab2e7dd9260130415c01a1f25561210", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 11, "status": "measured_file_manifest" }, "GGUF_BF16": { "bytes": 852896946208, "repo": "unsloth/MiniMax-M3-GGUF", "revision": "abdce34b5ab2e7dd9260130415c01a1f25561210", "url": "https://huggingface.co/unsloth/MiniMax-M3-GGUF/tree/abdce34b5ab2e7dd9260130415c01a1f25561210", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 18, "status": "measured_file_manifest" }, "MLX_4bit": { "bytes": 241481031547, "repo": "mlx-community/MiniMax-M3-4bit", "revision": "29909f4575501289881466f3b3bd910fdc95bc7c", "url": "https://huggingface.co/mlx-community/MiniMax-M3-4bit/tree/29909f4575501289881466f3b3bd910fdc95bc7c", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 58, "status": "measured_file_manifest", "mixed": true }, "MLX_8bit": { "bytes": 452576146772, "repo": "pipenetwork/MiniMax-M3-MLX-8bit", "revision": "9ba797b61fa3a33d7771f543cb0eb901abab785d", "url": "https://huggingface.co/pipenetwork/MiniMax-M3-MLX-8bit/tree/9ba797b61fa3a33d7771f543cb0eb901abab785d", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 87, "status": "measured_file_manifest", "mixed": true } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [], "notes": [] }, { "slug": "nex-n2-pro", "name": "Nex-N2-Pro (Qwen3.5-397B-A17B based)", "hf_repo": "nex-agi/Nex-N2-Pro", "hf_revision": "e7e4a4a29eb46a41a730d0b283ec5af0f16185d9", "source_url": "https://huggingface.co/nex-agi/Nex-N2-Pro", "org": "nex-agi", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-06", "precision": "month", "url": "https://github.com/nex-agi/Nex-N2/commits/main/README.md", "status": "verified_publisher_release_month", "note": "Publisher README release statement committed June3/4UTC; public Git availability date not proven. June9 is API promotion, not weight release." }, "params": { "total_b": 397, "active_b": 17, "moe": true, "url": "https://huggingface.co/nex-agi/Nex-N2-Pro/raw/e7e4a4a29eb46a41a730d0b283ec5af0f16185d9/README.md", "active_url": "https://huggingface.co/nex-agi/Nex-N2-Pro/raw/e7e4a4a29eb46a41a730d0b283ec5af0f16185d9/README.md", "scope": null }, "arch_label": "MoE · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/nex-agi/Nex-N2-Pro/raw/e7e4a4a29eb46a41a730d0b283ec5af0f16185d9/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/nex-agi/Nex-N2-Pro/blob/e7e4a4a29eb46a41a730d0b283ec5af0f16185d9/LICENSE" }, "quality": { "rank": 18, "aa": { "value": 28.1990972887912, "variant": "Nex-N2-Pro (Based on Qwen3.5-397B-A17B)", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/nex-n2-pro", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": 30720, "win_per_token": null, "win_tokens": null, "win_max": 0, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/nex-agi/Nex-N2-Pro/raw/e7e4a4a29eb46a41a730d0b283ec5af0f16185d9/config.json", "dtype": "BF16", "note": "BF16/FP16 KV estimate for full-attention layers only, batch=1. Sliding-window terms and fixed recurrent state are separate; excludes alignment, graphs, buffers, vision and MTP/drafters." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 241805932352, "repo": "bartowski/nex-agi_Nex-N2-Pro-GGUF", "revision": "f8b210fff0e6a75449a255a92ce88fc2fb170415", "url": "https://huggingface.co/bartowski/nex-agi_Nex-N2-Pro-GGUF/tree/f8b210fff0e6a75449a255a92ce88fc2fb170415", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 7, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-nex-agi_Nex-N2-Pro-bf16.gguf", "bytes": 921704480, "url": "https://huggingface.co/bartowski/nex-agi_Nex-N2-Pro-GGUF/resolve/f8b210fff0e6a75449a255a92ce88fc2fb170415/mmproj-nex-agi_Nex-N2-Pro-bf16.gguf" }, { "path": "mmproj-nex-agi_Nex-N2-Pro-f16.gguf", "bytes": 918165536, "url": "https://huggingface.co/bartowski/nex-agi_Nex-N2-Pro-GGUF/resolve/f8b210fff0e6a75449a255a92ce88fc2fb170415/mmproj-nex-agi_Nex-N2-Pro-f16.gguf" } ] }, "GGUF_Q5_K_M": { "bytes": 283257518080, "repo": "bartowski/nex-agi_Nex-N2-Pro-GGUF", "revision": "f8b210fff0e6a75449a255a92ce88fc2fb170415", "url": "https://huggingface.co/bartowski/nex-agi_Nex-N2-Pro-GGUF/tree/f8b210fff0e6a75449a255a92ce88fc2fb170415", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 8, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-nex-agi_Nex-N2-Pro-bf16.gguf", "bytes": 921704480, "url": "https://huggingface.co/bartowski/nex-agi_Nex-N2-Pro-GGUF/resolve/f8b210fff0e6a75449a255a92ce88fc2fb170415/mmproj-nex-agi_Nex-N2-Pro-bf16.gguf" }, { "path": "mmproj-nex-agi_Nex-N2-Pro-f16.gguf", "bytes": 918165536, "url": "https://huggingface.co/bartowski/nex-agi_Nex-N2-Pro-GGUF/resolve/f8b210fff0e6a75449a255a92ce88fc2fb170415/mmproj-nex-agi_Nex-N2-Pro-f16.gguf" } ] }, "GGUF_Q8_0": { "bytes": 421507364896, "repo": "bartowski/nex-agi_Nex-N2-Pro-GGUF", "revision": "f8b210fff0e6a75449a255a92ce88fc2fb170415", "url": "https://huggingface.co/bartowski/nex-agi_Nex-N2-Pro-GGUF/tree/f8b210fff0e6a75449a255a92ce88fc2fb170415", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 11, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-nex-agi_Nex-N2-Pro-bf16.gguf", "bytes": 921704480, "url": "https://huggingface.co/bartowski/nex-agi_Nex-N2-Pro-GGUF/resolve/f8b210fff0e6a75449a255a92ce88fc2fb170415/mmproj-nex-agi_Nex-N2-Pro-bf16.gguf" }, { "path": "mmproj-nex-agi_Nex-N2-Pro-f16.gguf", "bytes": 918165536, "url": "https://huggingface.co/bartowski/nex-agi_Nex-N2-Pro-GGUF/resolve/f8b210fff0e6a75449a255a92ce88fc2fb170415/mmproj-nex-agi_Nex-N2-Pro-f16.gguf" } ] }, "MLX_8bit": { "bytes": 421120909831, "repo": "mlx-community/Nex-N2-Pro-mlx-8bit", "revision": "5ad73cabde60f29697b128fa23130445ccc09aa8", "url": "https://huggingface.co/mlx-community/Nex-N2-Pro-mlx-8bit/tree/5ad73cabde60f29697b128fa23130445ccc09aa8", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 91, "status": "measured_file_manifest", "mixed": true } }, "artifacts_missing": [ "GGUF_FP16", "GGUF_BF16", "MLX_4bit" ], "community": [], "notes": [ "Derived from publisher explicitly naming Qwen3.5-397B-A17B as base; fine-tune tensor count is reported separately." ] }, { "slug": "nemotron-3-5-super-vl", "name": "NVIDIA Nemotron 3.5 Super VL 120B-A12B", "hf_repo": "nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16", "hf_revision": "1f8953f7f0935358e3605150d22bc15df062061c", "source_url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16", "org": "nvidia", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-10-15", "precision": "day", "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16/raw/main/README.md", "status": "verified_primary", "note": null }, "params": { "total_b": 120, "active_b": 12, "moe": true, "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16/raw/1f8953f7f0935358e3605150d22bc15df062061c/README.md", "active_url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16/raw/1f8953f7f0935358e3605150d22bc15df062061c/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16/raw/1f8953f7f0935358e3605150d22bc15df062061c/config.json", "extended": null }, "license": { "value": "openmdw-1.1", "url": "https://openmdw.ai/license/1-1/" }, "quality": { "rank": 19, "aa": { "value": 26.4910658661689, "variant": "Nemotron 3.5 Super VL", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/nemotron-3-5-super-vl", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1408.8993412571, "lower": 1397.1074604727, "upper": 1420.6912220414, "votes": 2523, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": 8192, "win_per_token": null, "win_tokens": null, "win_max": 0, "fixed": null, "unknown": [ "sliding_window", "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16/raw/1f8953f7f0935358e3605150d22bc15df062061c/config.json", "dtype": "BF16", "note": "Growing attention KV only. Mamba/recurrent/convolution states are extra and backend-specific. FFN/MoE blocks do not each add KV." }, "artifacts": { "GGUF_Q8_0": { "bytes": 131605437178, "repo": "unsloth/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-GGUF", "revision": "75a5829055f69a0359085ae6bbd0a81b60bc05c6", "url": "https://huggingface.co/unsloth/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-GGUF/tree/75a5829055f69a0359085ae6bbd0a81b60bc05c6", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 4, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 1733695840, "url": "https://huggingface.co/unsloth/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-GGUF/resolve/75a5829055f69a0359085ae6bbd0a81b60bc05c6/mmproj-BF16.gguf" } ] }, "MLX_4bit": { "bytes": 69414379784, "repo": "Duckets/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-MLX-4bit", "revision": "f171627c610b27f87c4ad9657638105a1dcc4d2e", "url": "https://huggingface.co/Duckets/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-MLX-4bit/tree/f171627c610b27f87c4ad9657638105a1dcc4d2e", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 14, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_FP16", "GGUF_BF16", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "kimi-k2.7-code", "name": "Kimi K2.7 Code", "hf_repo": "moonshotai/Kimi-K2.7-Code", "hf_revision": "74797c9c62378b951a1f6fcf5c4631024e9b8bef", "source_url": "https://huggingface.co/moonshotai/Kimi-K2.7-Code", "org": "moonshotai", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-06-12", "precision": "day", "url": "https://x.com/Kimi_Moonshot/status/2065377579130142937", "status": "publisher_statement_via_secondary_embed", "note": "Official open-source release statement embedded in https://gigazine.net/gsc_news/en/20260615-kimi-k2-7-code/; X inaccessible directly." }, "params": { "total_b": 1000, "active_b": 32, "moe": true, "url": "https://huggingface.co/moonshotai/Kimi-K2.7-Code/raw/74797c9c62378b951a1f6fcf5c4631024e9b8bef/README.md", "active_url": "https://huggingface.co/moonshotai/Kimi-K2.7-Code/raw/74797c9c62378b951a1f6fcf5c4631024e9b8bef/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 262144, "url": "https://huggingface.co/moonshotai/Kimi-K2.7-Code/raw/74797c9c62378b951a1f6fcf5c4631024e9b8bef/config.json", "extended": null }, "license": { "value": "modified-mit", "url": "https://huggingface.co/moonshotai/Kimi-K2.7-Code/blob/74797c9c62378b951a1f6fcf5c4631024e9b8bef/LICENSE" }, "quality": { "rank": 20, "aa": { "value": 25.8121062401836, "variant": "Kimi K2.7 Code", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/kimi-k2-7-code", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/moonshotai/Kimi-K2.7-Code/raw/74797c9c62378b951a1f6fcf5c4631024e9b8bef/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 620557904448, "repo": "DevQuasar/moonshotai.Kimi-K2.7-Code-GGUF", "revision": "b20b60b296b864df6283c5b37663e64b60e709e0", "url": "https://huggingface.co/DevQuasar/moonshotai.Kimi-K2.7-Code-GGUF/tree/b20b60b296b864df6283c5b37663e64b60e709e0", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 42, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-moonshotai.Kimi-K2.7-Code.f16.gguf", "bytes": 952572160, "url": "https://huggingface.co/DevQuasar/moonshotai.Kimi-K2.7-Code-GGUF/resolve/b20b60b296b864df6283c5b37663e64b60e709e0/mmproj-moonshotai.Kimi-K2.7-Code.f16.gguf" } ] }, "MLX_4bit": { "bytes": 641011788971, "repo": "mlx-community/Kimi-K2.7-Code-4bit", "revision": "8cae31c2e10dd628768ca3043c4eab6ca669edb3", "url": "https://huggingface.co/mlx-community/Kimi-K2.7-Code-4bit/tree/8cae31c2e10dd628768ca3043c4eab6ca669edb3", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 180, "status": "measured_file_manifest", "mixed": true } }, "artifacts_missing": [ "GGUF_Q5_K_M", "GGUF_Q8_0", "GGUF_FP16", "GGUF_BF16", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "inkling-small", "name": "Inkling-Small", "hf_repo": "thinkingmachines/Inkling-Small", "hf_revision": "8cc5877b44d343f88b92086aa1fb72897950f06a", "source_url": "https://huggingface.co/thinkingmachines/Inkling-Small", "org": "thinkingmachines", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-07-30", "precision": "day", "url": "https://thinkingmachines.ai/model-card/inkling-small/", "status": "verified_primary", "note": "July15 was smaller-model preview; July27 repository creation is not release." }, "params": { "total_b": 276, "active_b": 12, "moe": true, "url": "https://thinkingmachines.ai/news/inkling-small/", "active_url": "https://thinkingmachines.ai/news/inkling-small/", "scope": null }, "arch_label": "MoE · sliding-window attn", "context": { "native": 1048576, "url": "https://huggingface.co/thinkingmachines/Inkling-Small/raw/8cc5877b44d343f88b92086aa1fb72897950f06a/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://www.apache.org/licenses/LICENSE-2.0" }, "quality": { "rank": 21, "aa": { "value": 25.6582584411942, "variant": "Inkling Small", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/inkling-small", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1412.6234189083, "lower": 1407.9772596181, "upper": 1417.2695781985, "votes": 32304, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": 28672, "win_per_token": null, "win_tokens": null, "win_max": 73400320, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/thinkingmachines/Inkling-Small/raw/8cc5877b44d343f88b92086aa1fb72897950f06a/config.json", "dtype": "BF16", "note": "Backbone GQA/SWA KV only. Short-convolution state and optional8-layer MTP chain are extra; large Inkling uses16SWA KV heads versus8global KV heads." }, "artifacts": { "GGUF_Q8_0": { "bytes": 280278198720, "repo": "unsloth/Inkling-Small-GGUF", "revision": "1a19ef82883cb7b9c581b93c30ea252dabbf658d", "url": "https://huggingface.co/unsloth/Inkling-Small-GGUF/tree/1a19ef82883cb7b9c581b93c30ea252dabbf658d", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 7, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 138684128, "url": "https://huggingface.co/unsloth/Inkling-Small-GGUF/resolve/1a19ef82883cb7b9c581b93c30ea252dabbf658d/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 138684128, "url": "https://huggingface.co/unsloth/Inkling-Small-GGUF/resolve/1a19ef82883cb7b9c581b93c30ea252dabbf658d/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 277311968, "url": "https://huggingface.co/unsloth/Inkling-Small-GGUF/resolve/1a19ef82883cb7b9c581b93c30ea252dabbf658d/mmproj-F32.gguf" } ] }, "GGUF_BF16": { "bytes": 527412980416, "repo": "unsloth/Inkling-Small-GGUF", "revision": "1a19ef82883cb7b9c581b93c30ea252dabbf658d", "url": "https://huggingface.co/unsloth/Inkling-Small-GGUF/tree/1a19ef82883cb7b9c581b93c30ea252dabbf658d", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 11, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 138684128, "url": "https://huggingface.co/unsloth/Inkling-Small-GGUF/resolve/1a19ef82883cb7b9c581b93c30ea252dabbf658d/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 138684128, "url": "https://huggingface.co/unsloth/Inkling-Small-GGUF/resolve/1a19ef82883cb7b9c581b93c30ea252dabbf658d/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 277311968, "url": "https://huggingface.co/unsloth/Inkling-Small-GGUF/resolve/1a19ef82883cb7b9c581b93c30ea252dabbf658d/mmproj-F32.gguf" } ] }, "MLX_4bit": { "bytes": 153435815360, "repo": "mlx-community/Inkling-Small-mlx-4bit", "revision": "f0cafad5b1a3e54be06ba03fe07b4cd4e8bcc612", "url": "https://huggingface.co/mlx-community/Inkling-Small-mlx-4bit/tree/f0cafad5b1a3e54be06ba03fe07b4cd4e8bcc612", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 30, "status": "measured_file_manifest", "mixed": true } }, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_FP16", "MLX_8bit" ], "community": [], "notes": [ "Official publisher announcement says 276B; HF serialized tensor-element total differs." ] }, { "slug": "k2-horizon-mova-36b-a4b", "name": "K2-Horizon-MoVA-36B-A4B", "hf_repo": "IFM/K2-Horizon-MoVA-36B-A4B", "hf_revision": "e9ce717642b2074082875be233c02c729ba47960", "source_url": "https://huggingface.co/IFM/K2-Horizon-MoVA-36B-A4B", "org": "IFM", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-09-03", "precision": "day", "url": "https://mbzuai.ac.ae/news-events/news/mbzuais-institute-foundation-models-launches-k2-horizon-the-worlds-largest-fully", "status": "verified_primary", "note": "Publisher launch announces six models available immediately, including this size." }, "params": { "total_b": 36, "active_b": 4, "moe": true, "url": "https://huggingface.co/IFM/K2-Horizon-MoVA-36B-A4B/blob/e9ce717642b2074082875be233c02c729ba47960/README.md", "active_url": "https://huggingface.co/IFM/K2-Horizon-MoVA-36B-A4B/blob/e9ce717642b2074082875be233c02c729ba47960/README.md", "scope": "Nominal total/active publisher figures; MoVA attention needs custom cache accounting." }, "arch_label": "MoE", "context": { "native": 524288, "url": "https://huggingface.co/IFM/K2-Horizon-MoVA-36B-A4B/blob/e9ce717642b2074082875be233c02c729ba47960/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/IFM/K2-Horizon-MoVA-36B-A4B/blob/e9ce717642b2074082875be233c02c729ba47960/LICENSE" }, "quality": { "rank": 22, "aa": { "value": 25.3037790731435, "variant": "K2 Horizon MoVA 36B A4B", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/k2-horizon-mova-36b-a4b", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/IFM/K2-Horizon-MoVA-36B-A4B/blob/e9ce717642b2074082875be233c02c729ba47960/config.json", "dtype": "FP16", "note": "Architecture needs backend-specific treatment: shared K/V or cross-layer cache, looped layers, MoVA, MLA/KDA or diffusion. No naive all-layers GQA estimate is supplied." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 22368011616, "repo": "IFM/K2-Horizon-MoVA-36B-A4B-GGUF", "revision": "00cce3151bdce384661cb33603853ccfe6209de9", "url": "https://huggingface.co/IFM/K2-Horizon-MoVA-36B-A4B-GGUF/tree/00cce3151bdce384661cb33603853ccfe6209de9", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 26439456096, "repo": "IFM/K2-Horizon-MoVA-36B-A4B-GGUF", "revision": "00cce3151bdce384661cb33603853ccfe6209de9", "url": "https://huggingface.co/IFM/K2-Horizon-MoVA-36B-A4B-GGUF/tree/00cce3151bdce384661cb33603853ccfe6209de9", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 39831174496, "repo": "IFM/K2-Horizon-MoVA-36B-A4B-GGUF", "revision": "00cce3151bdce384661cb33603853ccfe6209de9", "url": "https://huggingface.co/IFM/K2-Horizon-MoVA-36B-A4B-GGUF/tree/00cce3151bdce384661cb33603853ccfe6209de9", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 26399514492, "repo": "mlx-community/K2-Horizon-MoVA-36B-A4B-4bit", "revision": "923353ec7a3eb0a85576ddb92e1496a753ea8e07", "url": "https://huggingface.co/mlx-community/K2-Horizon-MoVA-36B-A4B-4bit/tree/923353ec7a3eb0a85576ddb92e1496a753ea8e07", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 5, "status": "verified_hf_file_sizes", "mixed": true, "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 39803265096, "repo": "mlx-community/K2-Horizon-MoVA-36B-A4B-8bit", "revision": "f81c88b6a459e2f189db4dbac5f5fb3321559ccd", "url": "https://huggingface.co/mlx-community/K2-Horizon-MoVA-36B-A4B-8bit/tree/f81c88b6a459e2f189db4dbac5f5fb3321559ccd", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 8, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-10-02", "title": "This model with vision support would rock!", "url": "https://huggingface.co/IFM/K2-Horizon-MoVA-36B-A4B/discussions/10" }, { "platform": "Hugging Face", "date": "2026-09-10", "title": "Chat template for propper tool calling with llama.cpp?", "url": "https://huggingface.co/IFM/K2-Horizon-MoVA-36B-A4B/discussions/8" }, { "platform": "Reddit", "date": "2026-09-14", "title": null, "url": "https://www.reddit.com/r/LocalLLaMA/comments/1wg0vqz/the_new_k2_horizon_models_seem_like_an_absolute/" } ], "notes": [] }, { "slug": "hy3", "name": "Hy3", "hf_repo": "tencent/Hy3", "hf_revision": "a960ebc3da325ba167f069f76c41eb62c9280d22", "source_url": "https://huggingface.co/tencent/Hy3", "org": "tencent", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-07-06", "precision": "day", "url": "https://www.tencent.com/zh-hk/tencent-hunyuan-officially-releases-hy3-advancing-agent-capabilities-and-deeper-product-integration/", "status": "verified_primary", "note": "Publisher explicitly says Hugging Face weights available from day one." }, "params": { "total_b": 295, "active_b": 21, "moe": true, "url": "https://huggingface.co/tencent/Hy3/raw/a960ebc3da325ba167f069f76c41eb62c9280d22/README.md", "active_url": "https://huggingface.co/tencent/Hy3/raw/a960ebc3da325ba167f069f76c41eb62c9280d22/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 262144, "url": "https://huggingface.co/tencent/Hy3/raw/a960ebc3da325ba167f069f76c41eb62c9280d22/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/tencent/Hy3/blob/a960ebc3da325ba167f069f76c41eb62c9280d22/LICENSE" }, "quality": { "rank": 23, "aa": { "value": 25.2972775731571, "variant": "Hy3", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/hy3", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1438.0013487945, "lower": 1431.8690782021, "upper": 1444.1336193869, "votes": 12208, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": 327680, "win_per_token": null, "win_tokens": null, "win_max": 0, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/tencent/Hy3/raw/a960ebc3da325ba167f069f76c41eb62c9280d22/config.json", "dtype": "BF16", "note": "BF16/FP16 KV estimate for full-attention layers only, batch=1. Sliding-window terms and fixed recurrent state are separate; excludes alignment, graphs, buffers, vision and MTP/drafters." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 182328371648, "repo": "AngelSlim/Hy3-GGUF", "revision": "31b453f4d9b647c74e4c4f5cba632eb512332c91", "url": "https://huggingface.co/AngelSlim/Hy3-GGUF/tree/31b453f4d9b647c74e4c4f5cba632eb512332c91", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 1, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_Q5_K_M", "GGUF_Q8_0", "GGUF_FP16", "GGUF_BF16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "inkling", "name": "Inkling", "hf_repo": "thinkingmachines/Inkling", "hf_revision": "828496eeae4c243ff1a22f7f28ff83694f2f7bc9", "source_url": "https://huggingface.co/thinkingmachines/Inkling", "org": "thinkingmachines", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-07-15", "precision": "day", "url": "https://thinkingmachines.ai/news/introducing-inkling/", "status": "verified_primary", "note": null }, "params": { "total_b": 975, "active_b": 41, "moe": true, "url": "https://huggingface.co/thinkingmachines/Inkling/raw/828496eeae4c243ff1a22f7f28ff83694f2f7bc9/README.md", "active_url": "https://huggingface.co/thinkingmachines/Inkling/raw/828496eeae4c243ff1a22f7f28ff83694f2f7bc9/README.md", "scope": null }, "arch_label": "MoE · sliding-window attn", "context": { "native": 1048576, "url": "https://huggingface.co/thinkingmachines/Inkling/raw/828496eeae4c243ff1a22f7f28ff83694f2f7bc9/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://www.apache.org/licenses/LICENSE-2.0" }, "quality": { "rank": 24, "aa": { "value": 24.9847810999384, "variant": "Inkling (Xhigh)", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/inkling", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1441.3843755522, "lower": 1437.0850674699, "upper": 1445.6836836345, "votes": 40648, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": 45056, "win_per_token": null, "win_tokens": null, "win_max": 230686720, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/thinkingmachines/Inkling/raw/828496eeae4c243ff1a22f7f28ff83694f2f7bc9/config.json", "dtype": "BF16", "note": "Backbone GQA/SWA KV only. Short-convolution state and optional8-layer MTP chain are extra; large Inkling uses16SWA KV heads versus8global KV heads." }, "artifacts": { "GGUF_Q8_0": { "bytes": 856755583751, "repo": "unsloth/inkling-GGUF", "revision": "d3e9ffca48751dbe8b59dab5cfa364621257c682", "url": "https://huggingface.co/unsloth/inkling-GGUF/tree/d3e9ffca48751dbe8b59dab5cfa364621257c682", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 19, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 183264288, "url": "https://huggingface.co/unsloth/inkling-GGUF/resolve/d3e9ffca48751dbe8b59dab5cfa364621257c682/mmproj-BF16.gguf" } ] }, "GGUF_BF16": { "bytes": 1894278563239, "repo": "unsloth/inkling-GGUF", "revision": "d3e9ffca48751dbe8b59dab5cfa364621257c682", "url": "https://huggingface.co/unsloth/inkling-GGUF/tree/d3e9ffca48751dbe8b59dab5cfa364621257c682", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 41, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 183264288, "url": "https://huggingface.co/unsloth/inkling-GGUF/resolve/d3e9ffca48751dbe8b59dab5cfa364621257c682/mmproj-BF16.gguf" } ] }, "MLX_4bit": { "bytes": 548794689531, "repo": "mlx-community/Inkling-mlx-4bit", "revision": "1f2ad6fef3b19956d162c7995b4e63abfeaad565", "url": "https://huggingface.co/mlx-community/Inkling-mlx-4bit/tree/1f2ad6fef3b19956d162c7995b4e63abfeaad565", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 98, "status": "measured_file_manifest", "mixed": true } }, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_FP16", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "solar-open2-250b", "name": "Solar Open 2 (250B-A15B)", "hf_repo": "upstage/Solar-Open2-250B", "hf_revision": "9190fbe63a2ad8e17fc766ccceb36de7c66f004b", "source_url": "https://huggingface.co/upstage/Solar-Open2-250B", "org": "upstage", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-07-22", "precision": "day", "url": "https://www.upstage.ai/blog/en/solar-open-2", "status": "verified_primary", "note": "Official release blog date; AA August reflects evaluation/listing timing." }, "params": { "total_b": 250, "active_b": 15, "moe": true, "url": "https://huggingface.co/upstage/Solar-Open2-250B/raw/9190fbe63a2ad8e17fc766ccceb36de7c66f004b/README.md", "active_url": "https://huggingface.co/upstage/Solar-Open2-250B/raw/9190fbe63a2ad8e17fc766ccceb36de7c66f004b/README.md", "scope": null }, "arch_label": "MoE · hybrid linear attn", "context": { "native": 1048576, "url": "https://huggingface.co/upstage/Solar-Open2-250B/raw/9190fbe63a2ad8e17fc766ccceb36de7c66f004b/config.json", "extended": null }, "license": { "value": "upstage-solar-license", "url": "https://huggingface.co/upstage/Solar-Open2-250B/blob/9190fbe63a2ad8e17fc766ccceb36de7c66f004b/LICENSE" }, "quality": { "rank": 25, "aa": { "value": 24.7365297627523, "variant": "Solar Open2 250B", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/solar-open2-250b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": 49152, "win_per_token": null, "win_tokens": null, "win_max": 0, "fixed": null, "unknown": [ "sliding_window", "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/upstage/Solar-Open2-250B/raw/9190fbe63a2ad8e17fc766ccceb36de7c66f004b/config.json", "dtype": "BF16", "note": "Growing GQA KV only; linear attention convolution/recurrent state extra and backend-specific." }, "artifacts": { "GGUF_Q8_0": { "bytes": 266136847072, "repo": "Baekpica/Solar-Open2-250B-GGUF", "revision": "9a42599f82cf68886d063e1a76605996e4d9c1a7", "url": "https://huggingface.co/Baekpica/Solar-Open2-250B-GGUF/tree/9a42599f82cf68886d063e1a76605996e4d9c1a7", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 11, "status": "measured_file_manifest" }, "GGUF_BF16": { "bytes": 500718700672, "repo": "Baekpica/Solar-Open2-250B-GGUF", "revision": "9a42599f82cf68886d063e1a76605996e4d9c1a7", "url": "https://huggingface.co/Baekpica/Solar-Open2-250B-GGUF/tree/9a42599f82cf68886d063e1a76605996e4d9c1a7", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 11, "status": "measured_file_manifest" }, "MLX_4bit": { "bytes": 140883768096, "repo": "TensorFold/Solar-Open2-250B-MLX-4bit", "revision": "d34aa99f4f6f6a4c18e7aec3ece15a8a5c9809b8", "url": "https://huggingface.co/TensorFold/Solar-Open2-250B-MLX-4bit/tree/d34aa99f4f6f6a4c18e7aec3ece15a8a5c9809b8", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 29, "status": "measured_file_manifest" }, "MLX_8bit": { "bytes": 265994090275, "repo": "TensorFold/Solar-Open2-250B-MLX-8bit", "revision": "4b0e017be468e49c15b6176b4afc60e90ecb63de", "url": "https://huggingface.co/TensorFold/Solar-Open2-250B-MLX-8bit/tree/4b0e017be468e49c15b6176b4afc60e90ecb63de", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 62, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_FP16" ], "community": [], "notes": [] }, { "slug": "ling-3-0-flash-vl", "name": "Ling-3.0-flash-VL", "hf_repo": "inclusionAI/Ling-3.0-flash-VL", "hf_revision": "c22bf252f15b66d18ca00a15989ac7ed46b7f09e", "source_url": "https://huggingface.co/inclusionAI/Ling-3.0-flash-VL", "org": "inclusionAI", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-09-09", "precision": "day", "url": "https://www.ant-ling.com/en", "status": "verified_primary", "note": "Publisher explicitly labels open-source release; distinguish earlier hosted rollout." }, "params": { "total_b": 124, "active_b": 5.5, "moe": true, "url": "https://huggingface.co/inclusionAI/Ling-3.0-flash-VL/raw/c22bf252f15b66d18ca00a15989ac7ed46b7f09e/README.md", "active_url": "https://huggingface.co/inclusionAI/Ling-3.0-flash-VL/raw/c22bf252f15b66d18ca00a15989ac7ed46b7f09e/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 131072, "url": "https://huggingface.co/inclusionAI/Ling-3.0-flash-VL/raw/c22bf252f15b66d18ca00a15989ac7ed46b7f09e/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/inclusionAI/Ling-3.0-flash-VL/blob/c22bf252f15b66d18ca00a15989ac7ed46b7f09e/LICENSE" }, "quality": { "rank": 26, "aa": { "value": 24.5693054077875, "variant": "Ling-3.0-flash-VL", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/ling-3-0-flash-vl", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/inclusionAI/Ling-3.0-flash-VL/raw/c22bf252f15b66d18ca00a15989ac7ed46b7f09e/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 78661520544, "repo": "bartowski/Ling-3.0-flash-VL-GGUF", "revision": "2904b2b35bf3fde46a9f6740c92939aa32a5ba28", "url": "https://huggingface.co/bartowski/Ling-3.0-flash-VL-GGUF/tree/2904b2b35bf3fde46a9f6740c92939aa32a5ba28", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 2, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-Ling-3.0-flash-VL-bf16.gguf", "bytes": 878174272, "url": "https://huggingface.co/bartowski/Ling-3.0-flash-VL-GGUF/resolve/2904b2b35bf3fde46a9f6740c92939aa32a5ba28/mmproj-Ling-3.0-flash-VL-bf16.gguf" }, { "path": "mmproj-Ling-3.0-flash-VL-f16.gguf", "bytes": 874635328, "url": "https://huggingface.co/bartowski/Ling-3.0-flash-VL-GGUF/resolve/2904b2b35bf3fde46a9f6740c92939aa32a5ba28/mmproj-Ling-3.0-flash-VL-f16.gguf" } ] }, "GGUF_Q5_K_M": { "bytes": 95884381472, "repo": "bartowski/Ling-3.0-flash-VL-GGUF", "revision": "2904b2b35bf3fde46a9f6740c92939aa32a5ba28", "url": "https://huggingface.co/bartowski/Ling-3.0-flash-VL-GGUF/tree/2904b2b35bf3fde46a9f6740c92939aa32a5ba28", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 3, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-Ling-3.0-flash-VL-bf16.gguf", "bytes": 878174272, "url": "https://huggingface.co/bartowski/Ling-3.0-flash-VL-GGUF/resolve/2904b2b35bf3fde46a9f6740c92939aa32a5ba28/mmproj-Ling-3.0-flash-VL-bf16.gguf" }, { "path": "mmproj-Ling-3.0-flash-VL-f16.gguf", "bytes": 874635328, "url": "https://huggingface.co/bartowski/Ling-3.0-flash-VL-GGUF/resolve/2904b2b35bf3fde46a9f6740c92939aa32a5ba28/mmproj-Ling-3.0-flash-VL-f16.gguf" } ] }, "GGUF_Q8_0": { "bytes": 132365277248, "repo": "bartowski/Ling-3.0-flash-VL-GGUF", "revision": "2904b2b35bf3fde46a9f6740c92939aa32a5ba28", "url": "https://huggingface.co/bartowski/Ling-3.0-flash-VL-GGUF/tree/2904b2b35bf3fde46a9f6740c92939aa32a5ba28", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 4, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-Ling-3.0-flash-VL-bf16.gguf", "bytes": 878174272, "url": "https://huggingface.co/bartowski/Ling-3.0-flash-VL-GGUF/resolve/2904b2b35bf3fde46a9f6740c92939aa32a5ba28/mmproj-Ling-3.0-flash-VL-bf16.gguf" }, { "path": "mmproj-Ling-3.0-flash-VL-f16.gguf", "bytes": 874635328, "url": "https://huggingface.co/bartowski/Ling-3.0-flash-VL-GGUF/resolve/2904b2b35bf3fde46a9f6740c92939aa32a5ba28/mmproj-Ling-3.0-flash-VL-f16.gguf" } ] }, "GGUF_BF16": { "bytes": 248944047584, "repo": "bartowski/Ling-3.0-flash-VL-GGUF", "revision": "2904b2b35bf3fde46a9f6740c92939aa32a5ba28", "url": "https://huggingface.co/bartowski/Ling-3.0-flash-VL-GGUF/tree/2904b2b35bf3fde46a9f6740c92939aa32a5ba28", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 7, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-Ling-3.0-flash-VL-bf16.gguf", "bytes": 878174272, "url": "https://huggingface.co/bartowski/Ling-3.0-flash-VL-GGUF/resolve/2904b2b35bf3fde46a9f6740c92939aa32a5ba28/mmproj-Ling-3.0-flash-VL-bf16.gguf" }, { "path": "mmproj-Ling-3.0-flash-VL-f16.gguf", "bytes": 874635328, "url": "https://huggingface.co/bartowski/Ling-3.0-flash-VL-GGUF/resolve/2904b2b35bf3fde46a9f6740c92939aa32a5ba28/mmproj-Ling-3.0-flash-VL-f16.gguf" } ] } }, "artifacts_missing": [ "GGUF_FP16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "nemotron-3-ultra-550b-a55b", "name": "NVIDIA Nemotron 3 Ultra 550B-A55B", "hf_repo": "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16", "hf_revision": "77df655d5e9f8362164ed14dd8b48f8bce657498", "source_url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16", "org": "nvidia", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-06-04", "precision": "day", "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16/raw/main/README.md", "status": "verified_primary", "note": null }, "params": { "total_b": 550, "active_b": 55, "moe": true, "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16/raw/77df655d5e9f8362164ed14dd8b48f8bce657498/README.md", "active_url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16/raw/77df655d5e9f8362164ed14dd8b48f8bce657498/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 262144, "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16/raw/77df655d5e9f8362164ed14dd8b48f8bce657498/config.json", "extended": null }, "license": { "value": "openmdw-1.1", "url": "https://openmdw.ai/license/1-1/" }, "quality": { "rank": 27, "aa": { "value": 22.9279123422858, "variant": "Nemotron 3 Ultra 550B A55B (Reasoning)", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/nvidia-nemotron-3-ultra-550b-a55b", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1444.6352226175, "lower": 1437.7820965364, "upper": 1451.4883486986, "votes": 11567, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": 12288, "win_per_token": null, "win_tokens": null, "win_max": 0, "fixed": null, "unknown": [ "sliding_window", "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16/raw/77df655d5e9f8362164ed14dd8b48f8bce657498/config.json", "dtype": "BF16", "note": "Growing attention KV only. Mamba/recurrent/convolution states are extra and backend-specific. FFN/MoE blocks do not each add KV." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 345131239808, "repo": "DevQuasar/nvidia.NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16-GGUF", "revision": "6eb096e1b3da0dad78d937efa6de8ee742a2410e", "url": "https://huggingface.co/DevQuasar/nvidia.NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16-GGUF/tree/6eb096e1b3da0dad78d937efa6de8ee742a2410e", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 24, "status": "measured_file_manifest" }, "GGUF_Q5_K_M": { "bytes": 396593777664, "repo": "DevQuasar/nvidia.NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16-GGUF", "revision": "6eb096e1b3da0dad78d937efa6de8ee742a2410e", "url": "https://huggingface.co/DevQuasar/nvidia.NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16-GGUF/tree/6eb096e1b3da0dad78d937efa6de8ee742a2410e", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 30, "status": "measured_file_manifest" }, "GGUF_Q8_0": { "bytes": 584258241056, "repo": "unsloth/NVIDIA-Nemotron-3-Ultra-550B-A55B-GGUF", "revision": "2fb7d5b3f4eae7aedb18b4839b6a6300111e46f6", "url": "https://huggingface.co/unsloth/NVIDIA-Nemotron-3-Ultra-550B-A55B-GGUF/tree/2fb7d5b3f4eae7aedb18b4839b6a6300111e46f6", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 14, "status": "measured_file_manifest" }, "GGUF_BF16": { "bytes": 1099040900576, "repo": "unsloth/NVIDIA-Nemotron-3-Ultra-550B-A55B-GGUF", "revision": "2fb7d5b3f4eae7aedb18b4839b6a6300111e46f6", "url": "https://huggingface.co/unsloth/NVIDIA-Nemotron-3-Ultra-550B-A55B-GGUF/tree/2fb7d5b3f4eae7aedb18b4839b6a6300111e46f6", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 24, "status": "measured_file_manifest" }, "MLX_4bit": { "bytes": 309284841338, "repo": "pipenetwork/NVIDIA-Nemotron-3-Ultra-550B-A55B-MLX-4bit", "revision": "d64007dfeb7eb87eba579f34fdb03cf329077fd8", "url": "https://huggingface.co/pipenetwork/NVIDIA-Nemotron-3-Ultra-550B-A55B-MLX-4bit/tree/d64007dfeb7eb87eba579f34fdb03cf329077fd8", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 96, "status": "measured_file_manifest" }, "MLX_8bit": { "bytes": 583835595700, "repo": "pipenetwork/NVIDIA-Nemotron-3-Ultra-550B-A55B-MLX-8bit", "revision": "e338816a1d3c11ac6601bbd2cdf35500ce941fda", "url": "https://huggingface.co/pipenetwork/NVIDIA-Nemotron-3-Ultra-550B-A55B-MLX-8bit/tree/e338816a1d3c11ac6601bbd2cdf35500ce941fda", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 193, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [], "notes": [] }, { "slug": "ling-3-0-flash-fin", "name": "Ling-3.0-flash-Fin", "hf_repo": "inclusionAI/Ling-3.0-flash-Fin", "hf_revision": "a5b4afb48f526175991724472c179e5a065604de", "source_url": "https://huggingface.co/inclusionAI/Ling-3.0-flash-Fin", "org": "inclusionAI", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026", "precision": "year", "url": "https://www.ant-ling.com/en", "status": "derived_from_publisher_base_release_and_live_weight_availability", "note": "Fin is a later fine-tune of Ling3.0Flash first published in2026; exact fin release day unverified." }, "params": { "total_b": 124, "active_b": 5.1, "moe": true, "url": "https://huggingface.co/inclusionAI/Ling-3.0-flash-Fin/raw/a5b4afb48f526175991724472c179e5a065604de/README.md", "active_url": "https://huggingface.co/inclusionAI/Ling-3.0-flash-Fin/raw/a5b4afb48f526175991724472c179e5a065604de/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 262144, "url": "https://huggingface.co/inclusionAI/Ling-3.0-flash-Fin/raw/a5b4afb48f526175991724472c179e5a065604de/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/inclusionAI/Ling-3.0-flash-Fin/blob/a5b4afb48f526175991724472c179e5a065604de/LICENSE" }, "quality": { "rank": 28, "aa": { "value": 22.5965828940594, "variant": "Ling-3.0-flash-Fin", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/ling-3-0-flash-fin", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/inclusionAI/Ling-3.0-flash-Fin/raw/a5b4afb48f526175991724472c179e5a065604de/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 77804990560, "repo": "bartowski/Ling-3.0-flash-Fin-GGUF", "revision": "a792b6bd51d2cb47b5aa8687ff5dd4f2974acdb4", "url": "https://huggingface.co/bartowski/Ling-3.0-flash-Fin-GGUF/tree/a792b6bd51d2cb47b5aa8687ff5dd4f2974acdb4", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 2, "status": "measured_file_manifest" }, "GGUF_Q5_K_M": { "bytes": 90661271776, "repo": "bartowski/Ling-3.0-flash-Fin-GGUF", "revision": "a792b6bd51d2cb47b5aa8687ff5dd4f2974acdb4", "url": "https://huggingface.co/bartowski/Ling-3.0-flash-Fin-GGUF/tree/a792b6bd51d2cb47b5aa8687ff5dd4f2974acdb4", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 3, "status": "measured_file_manifest" }, "GGUF_Q8_0": { "bytes": 135633373216, "repo": "bartowski/Ling-3.0-flash-Fin-GGUF", "revision": "a792b6bd51d2cb47b5aa8687ff5dd4f2974acdb4", "url": "https://huggingface.co/bartowski/Ling-3.0-flash-Fin-GGUF/tree/a792b6bd51d2cb47b5aa8687ff5dd4f2974acdb4", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 4, "status": "measured_file_manifest" }, "GGUF_BF16": { "bytes": 255091083616, "repo": "bartowski/Ling-3.0-flash-Fin-GGUF", "revision": "a792b6bd51d2cb47b5aa8687ff5dd4f2974acdb4", "url": "https://huggingface.co/bartowski/Ling-3.0-flash-Fin-GGUF/tree/a792b6bd51d2cb47b5aa8687ff5dd4f2974acdb4", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 7, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_FP16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "g9v3-39a5b", "name": "G9v3-39A5B (preview)", "hf_repo": "ai9stars/G9v3-39A5B", "hf_revision": "70213b14ea559298507e46e36e53d9c4c76144fc", "source_url": "https://huggingface.co/ai9stars/G9v3-39A5B", "org": "ai9stars", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08", "precision": "month", "url": "https://artificialanalysis.ai/models/g9v3-39a5b", "status": "verified_secondary_aa_release_metadata", "note": "AA explicitly reports release month; exact primary-source public release day remains unknown." }, "params": { "total_b": 39, "active_b": 5, "moe": true, "url": "https://huggingface.co/ai9stars/G9v3-39A5B/blob/70213b14ea559298507e46e36e53d9c4c76144fc/README.md", "active_url": "https://huggingface.co/ai9stars/G9v3-39A5B/blob/70213b14ea559298507e46e36e53d9c4c76144fc/README.md", "scope": "Publisher approximate total/active counts; preview." }, "arch_label": "MoE", "context": { "native": 131072, "url": "https://huggingface.co/ai9stars/G9v3-39A5B/blob/70213b14ea559298507e46e36e53d9c4c76144fc/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/ai9stars/G9v3-39A5B/blob/70213b14ea559298507e46e36e53d9c4c76144fc/LICENSE" }, "quality": { "rank": 29, "aa": { "value": 21.8316172646825, "variant": "G9v3-39A5B", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/g9v3-39a5b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": 38912, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/ai9stars/G9v3-39A5B/blob/70213b14ea559298507e46e36e53d9c4c76144fc/config.json", "dtype": "FP16", "note": "Ordinary GQA cache estimate; actual allocation and padding vary by backend." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 23559222240, "repo": "linuxid10t/G9v3-39A5B-GGUF", "revision": "a6956c38a55abdbc22a120c51f735cd87af17b61", "url": "https://huggingface.co/linuxid10t/G9v3-39A5B-GGUF/tree/a6956c38a55abdbc22a120c51f735cd87af17b61", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 41479836640, "repo": "linuxid10t/G9v3-39A5B-GGUF", "revision": "a6956c38a55abdbc22a120c51f735cd87af17b61", "url": "https://huggingface.co/linuxid10t/G9v3-39A5B-GGUF/tree/a6956c38a55abdbc22a120c51f735cd87af17b61", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 22803310221, "repo": "rapid-mlx/G9v3-39A5B-MLX-4bit", "revision": "6a5ec618f742d189fd1fcf3bd86aed167465aec4", "url": "https://huggingface.co/rapid-mlx/G9v3-39A5B-MLX-4bit/tree/6a5ec618f742d189fd1fcf3bd86aed167465aec4", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 5, "status": "verified_hf_file_sizes", "mixed": true, "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_Q5_K_M", "GGUF_FP16", "MLX_8bit" ], "community": [ { "platform": "Hugging Face", "date": "2026-08-04", "title": "GGUF Version?", "url": "https://huggingface.co/ai9stars/G9v3-39A5B/discussions/3" }, { "platform": "Hugging Face", "date": "2026-08-20", "title": "Is model MTP trained?", "url": "https://huggingface.co/ai9stars/G9v3-39A5B/discussions/4" } ], "notes": [ "Exact public release day not verified from a primary announcement; repository creation and commit dates are deliberately not substituted." ] }, { "slug": "a-x-k2", "name": "A.X K2 (SK Telecom)", "hf_repo": "skt/A.X-K2", "hf_revision": "2287ca456927eed33b899c404c0b2ffaf17aa09f", "source_url": "https://huggingface.co/skt/A.X-K2", "org": "skt", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-07-29", "precision": "day", "url": "https://news.sktelecom.com/en/3204", "status": "verified_primary", "note": "Publisher explicitly says weights on Hugging Face starting today." }, "params": { "total_b": 688, "active_b": 33, "moe": true, "url": "https://huggingface.co/skt/A.X-K2/raw/2287ca456927eed33b899c404c0b2ffaf17aa09f/README.md", "active_url": "https://huggingface.co/skt/A.X-K2/raw/2287ca456927eed33b899c404c0b2ffaf17aa09f/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 262144, "url": "https://huggingface.co/skt/A.X-K2/raw/2287ca456927eed33b899c404c0b2ffaf17aa09f/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/skt/A.X-K2/blob/2287ca456927eed33b899c404c0b2ffaf17aa09f/LICENSE" }, "quality": { "rank": 30, "aa": { "value": 21.4612645385074, "variant": "A.X-K2", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/a-x-k2", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/skt/A.X-K2/raw/2287ca456927eed33b899c404c0b2ffaf17aa09f/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": {}, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_Q8_0", "GGUF_FP16", "GGUF_BF16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "qwen3.5-397b-a17b", "name": "Qwen3.5-397B-A17B", "hf_repo": "Qwen/Qwen3.5-397B-A17B", "hf_revision": "8472618112abcbd45acbcdc58436aff4233c23f7", "source_url": "https://huggingface.co/Qwen/Qwen3.5-397B-A17B", "org": "Qwen", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-02-16", "precision": "day", "url": "https://github.com/QwenLM/Qwen3.8", "status": "verified_primary", "note": null }, "params": { "total_b": 397, "active_b": 17, "moe": true, "url": "https://huggingface.co/Qwen/Qwen3.5-397B-A17B/raw/8472618112abcbd45acbcdc58436aff4233c23f7/README.md", "active_url": "https://huggingface.co/Qwen/Qwen3.5-397B-A17B/raw/8472618112abcbd45acbcdc58436aff4233c23f7/README.md", "scope": null }, "arch_label": "MoE · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/Qwen/Qwen3.5-397B-A17B/raw/8472618112abcbd45acbcdc58436aff4233c23f7/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/Qwen/Qwen3.5-397B-A17B/blob/main/LICENSE" }, "quality": { "rank": 31, "aa": { "value": 21.4366547759392, "variant": "Qwen3.5 397B A17B (Non-reasoning)", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/qwen3-5-397b-a17b-non-reasoning", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": { "value": 1437.7189287177, "lower": 1434.6508668877, "upper": 1440.7869905477, "votes": 93567, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": 30720, "win_per_token": null, "win_tokens": null, "win_max": 0, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3.5-397B-A17B/raw/8472618112abcbd45acbcdc58436aff4233c23f7/config.json", "dtype": "BF16", "note": "BF16/FP16 KV estimate for full-attention layers only, batch=1. Sliding-window terms and fixed recurrent state are separate; excludes alignment, graphs, buffers, vision and MTP/drafters." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 244093630912, "repo": "unsloth/Qwen3.5-397B-A17B-GGUF", "revision": "da33c16fa4440f831149fcf53b98a22bc07785e5", "url": "https://huggingface.co/unsloth/Qwen3.5-397B-A17B-GGUF/tree/da33c16fa4440f831149fcf53b98a22bc07785e5", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 6, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 921705184, "url": "https://huggingface.co/unsloth/Qwen3.5-397B-A17B-GGUF/resolve/da33c16fa4440f831149fcf53b98a22bc07785e5/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 918166240, "url": "https://huggingface.co/unsloth/Qwen3.5-397B-A17B-GGUF/resolve/da33c16fa4440f831149fcf53b98a22bc07785e5/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1824062176, "url": "https://huggingface.co/unsloth/Qwen3.5-397B-A17B-GGUF/resolve/da33c16fa4440f831149fcf53b98a22bc07785e5/mmproj-F32.gguf" } ] }, "GGUF_Q5_K_M": { "bytes": 293664979584, "repo": "unsloth/Qwen3.5-397B-A17B-GGUF", "revision": "da33c16fa4440f831149fcf53b98a22bc07785e5", "url": "https://huggingface.co/unsloth/Qwen3.5-397B-A17B-GGUF/tree/da33c16fa4440f831149fcf53b98a22bc07785e5", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 8, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 921705184, "url": "https://huggingface.co/unsloth/Qwen3.5-397B-A17B-GGUF/resolve/da33c16fa4440f831149fcf53b98a22bc07785e5/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 918166240, "url": "https://huggingface.co/unsloth/Qwen3.5-397B-A17B-GGUF/resolve/da33c16fa4440f831149fcf53b98a22bc07785e5/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1824062176, "url": "https://huggingface.co/unsloth/Qwen3.5-397B-A17B-GGUF/resolve/da33c16fa4440f831149fcf53b98a22bc07785e5/mmproj-F32.gguf" } ] }, "GGUF_Q8_0": { "bytes": 421507365824, "repo": "unsloth/Qwen3.5-397B-A17B-GGUF", "revision": "da33c16fa4440f831149fcf53b98a22bc07785e5", "url": "https://huggingface.co/unsloth/Qwen3.5-397B-A17B-GGUF/tree/da33c16fa4440f831149fcf53b98a22bc07785e5", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 10, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 921705184, "url": "https://huggingface.co/unsloth/Qwen3.5-397B-A17B-GGUF/resolve/da33c16fa4440f831149fcf53b98a22bc07785e5/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 918166240, "url": "https://huggingface.co/unsloth/Qwen3.5-397B-A17B-GGUF/resolve/da33c16fa4440f831149fcf53b98a22bc07785e5/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1824062176, "url": "https://huggingface.co/unsloth/Qwen3.5-397B-A17B-GGUF/resolve/da33c16fa4440f831149fcf53b98a22bc07785e5/mmproj-F32.gguf" } ] }, "GGUF_BF16": { "bytes": 792961318816, "repo": "unsloth/Qwen3.5-397B-A17B-GGUF", "revision": "da33c16fa4440f831149fcf53b98a22bc07785e5", "url": "https://huggingface.co/unsloth/Qwen3.5-397B-A17B-GGUF/tree/da33c16fa4440f831149fcf53b98a22bc07785e5", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 17, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 921705184, "url": "https://huggingface.co/unsloth/Qwen3.5-397B-A17B-GGUF/resolve/da33c16fa4440f831149fcf53b98a22bc07785e5/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 918166240, "url": "https://huggingface.co/unsloth/Qwen3.5-397B-A17B-GGUF/resolve/da33c16fa4440f831149fcf53b98a22bc07785e5/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1824062176, "url": "https://huggingface.co/unsloth/Qwen3.5-397B-A17B-GGUF/resolve/da33c16fa4440f831149fcf53b98a22bc07785e5/mmproj-F32.gguf" } ] }, "MLX_4bit": { "bytes": 223861162211, "repo": "mlx-community/Qwen3.5-397B-A17B-4bit", "revision": "39159bd8aa74f5c8446d2b2dc584f62bb51cb0d3", "url": "https://huggingface.co/mlx-community/Qwen3.5-397B-A17B-4bit/tree/39159bd8aa74f5c8446d2b2dc584f62bb51cb0d3", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 46, "status": "measured_file_manifest" }, "MLX_8bit": { "bytes": 422032974222, "repo": "mlx-community/Qwen3.5-397B-A17B-8bit", "revision": "ab5601e0191bf1f625b48453a90b73bef66482a6", "url": "https://huggingface.co/mlx-community/Qwen3.5-397B-A17B-8bit/tree/ab5601e0191bf1f625b48453a90b73bef66482a6", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 91, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [], "notes": [] }, { "slug": "nemotron-3-5-bolt-74b-a9b", "name": "NVIDIA Nemotron 3.5 Bolt 74B-A9B", "hf_repo": "nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-BF16", "hf_revision": "a6a1e5030819a130a1ea08aadb1fe42f4cc86777", "source_url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-BF16", "org": "nvidia", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-10-15", "precision": "day", "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-BF16/raw/main/README.md", "status": "verified_primary", "note": null }, "params": { "total_b": 74, "active_b": 9, "moe": true, "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-BF16/raw/a6a1e5030819a130a1ea08aadb1fe42f4cc86777/README.md", "active_url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-BF16/raw/a6a1e5030819a130a1ea08aadb1fe42f4cc86777/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 262144, "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-BF16/raw/a6a1e5030819a130a1ea08aadb1fe42f4cc86777/config.json", "extended": null }, "license": { "value": "openmdw-1.1", "url": "https://openmdw.ai/license/1-1/" }, "quality": { "rank": 32, "aa": { "value": 21.4072054842003, "variant": "Nemotron 3.5 Bolt 74B-A9B", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/nemotron-3-5-bolt-74b-a9b", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": 8192, "win_per_token": null, "win_tokens": null, "win_max": 0, "fixed": null, "unknown": [ "sliding_window", "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-BF16/raw/a6a1e5030819a130a1ea08aadb1fe42f4cc86777/config.json", "dtype": "BF16", "note": "Growing attention KV only. Mamba/recurrent/convolution states are extra and backend-specific. FFN/MoE blocks do not each add KV." }, "artifacts": { "GGUF_Q8_0": { "bytes": 78852779040, "repo": "unsloth/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-GGUF", "revision": "34841e360524d421c564cc3ed247aa04a865473b", "url": "https://huggingface.co/unsloth/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-GGUF/tree/34841e360524d421c564cc3ed247aa04a865473b", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 3, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 1733692672, "url": "https://huggingface.co/unsloth/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-GGUF/resolve/34841e360524d421c564cc3ed247aa04a865473b/mmproj-BF16.gguf" } ] }, "GGUF_BF16": { "bytes": 148212149408, "repo": "unsloth/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-GGUF", "revision": "34841e360524d421c564cc3ed247aa04a865473b", "url": "https://huggingface.co/unsloth/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-GGUF/tree/34841e360524d421c564cc3ed247aa04a865473b", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 4, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 1733692672, "url": "https://huggingface.co/unsloth/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-GGUF/resolve/34841e360524d421c564cc3ed247aa04a865473b/mmproj-BF16.gguf" } ] }, "MLX_4bit": { "bytes": 43257123023, "repo": "rapid-mlx/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-MLX-4bit", "revision": "cbfda6be4e9ff9e8d3acf0c0a57c54c1a496ef81", "url": "https://huggingface.co/rapid-mlx/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-MLX-4bit/tree/cbfda6be4e9ff9e8d3acf0c0a57c54c1a496ef81", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 9, "status": "measured_file_manifest", "mixed": true }, "MLX_8bit": { "bytes": 80234120732, "repo": "nativ-community/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-8bit", "revision": "fe2ac821168e14ac90110bf7d4cba4b61078a30d", "url": "https://huggingface.co/nativ-community/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-8bit/tree/fe2ac821168e14ac90110bf7d4cba4b61078a30d", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 17, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_FP16" ], "community": [], "notes": [ "Official 24GB marketing label is not a measured total runtime RAM claim. Main+projector+optional drafter weights total24,750,261,824bytes (23.05GiB), before attention KV, Mamba state, graphs and backend buffers." ] }, { "slug": "k2-horizon-7b", "name": "K2-Horizon-7B", "hf_repo": "IFM/K2-Horizon-7B", "hf_revision": "4622f94a6632384784471fd674d5b78f8d527798", "source_url": "https://huggingface.co/IFM/K2-Horizon-7B", "org": "IFM", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-09-03", "precision": "day", "url": "https://mbzuai.ac.ae/news-events/news/mbzuais-institute-foundation-models-launches-k2-horizon-the-worlds-largest-fully", "status": "verified_primary", "note": "Publisher launch announces six models available immediately, including this size." }, "params": { "total_b": null, "active_b": null, "moe": false, "url": null, "active_url": null, "scope": "7B is core/non-embedding class; HF stored tensor count includes large embeddings." }, "arch_label": "Dense", "context": { "native": 524288, "url": "https://huggingface.co/IFM/K2-Horizon-7B/blob/4622f94a6632384784471fd674d5b78f8d527798/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/IFM/K2-Horizon-7B/blob/4622f94a6632384784471fd674d5b78f8d527798/LICENSE" }, "quality": { "rank": 33, "aa": { "value": 20.5958819421354, "variant": "K2 Horizon 7B", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/k2-horizon-7b", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": 147456, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/IFM/K2-Horizon-7B/blob/4622f94a6632384784471fd674d5b78f8d527798/config.json", "dtype": "FP16", "note": "Ordinary GQA cache estimate; actual allocation and padding vary by backend." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 5592217984, "repo": "IFM/K2-Horizon-7B-GGUF", "revision": "7a75dbda05e3d440351543131ff3f7fa5687e879", "url": "https://huggingface.co/IFM/K2-Horizon-7B-GGUF/tree/7a75dbda05e3d440351543131ff3f7fa5687e879", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 6466075008, "repo": "IFM/K2-Horizon-7B-GGUF", "revision": "7a75dbda05e3d440351543131ff3f7fa5687e879", "url": "https://huggingface.co/IFM/K2-Horizon-7B-GGUF/tree/7a75dbda05e3d440351543131ff3f7fa5687e879", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 9573964160, "repo": "IFM/K2-Horizon-7B-GGUF", "revision": "7a75dbda05e3d440351543131ff3f7fa5687e879", "url": "https://huggingface.co/IFM/K2-Horizon-7B-GGUF/tree/7a75dbda05e3d440351543131ff3f7fa5687e879", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 18010422688, "repo": "mradermacher/K2-Horizon-7B-GGUF", "revision": "d7fc21eb33d3058b3a2debb4c95abeed4b20febb", "url": "https://huggingface.co/mradermacher/K2-Horizon-7B-GGUF/tree/d7fc21eb33d3058b3a2debb4c95abeed4b20febb", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 7183832504, "repo": "mlx-community/K2-Horizon-7B-4bit", "revision": "c8698376a102bc5e7ad49089b21b62c096a2506c", "url": "https://huggingface.co/mlx-community/K2-Horizon-7B-4bit/tree/c8698376a102bc5e7ad49089b21b62c096a2506c", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 2, "status": "verified_hf_file_sizes", "mixed": true, "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 9562002650, "repo": "mlx-community/K2-Horizon-7B-8bit", "revision": "713b027d9416161ef6d03e50691b523aed66acea", "url": "https://huggingface.co/mlx-community/K2-Horizon-7B-8bit/tree/713b027d9416161ef6d03e50691b523aed66acea", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 2, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-10-11", "title": "Vision?", "url": "https://huggingface.co/IFM/K2-Horizon-7B/discussions/8" }, { "platform": "Hugging Face", "date": "2026-10-05", "title": "repo size is 18B and seems like 9B model not 7B", "url": "https://huggingface.co/IFM/K2-Horizon-7B/discussions/6" }, { "platform": "Reddit", "date": "2026-09-15", "title": null, "url": "https://www.reddit.com/r/LocalLLaMA/comments/1wha6sv/i_benchmarked_ifmk2horizon7b_on_16gb_vram/" }, { "platform": "Reddit", "date": "2026-09-14", "title": null, "url": "https://www.reddit.com/r/LocalLLaMA/comments/1wg0vqz/the_new_k2_horizon_models_seem_like_an_absolute/" } ], "notes": [] }, { "slug": "k-exaone-2.0-750b-a37b", "name": "K-EXAONE 2.0 750B-A37B", "hf_repo": "LGAI-EXAONE/K-EXAONE-2.0-750B-A37B", "hf_revision": "fec0c8dd25fdd20ffd09869d5af5277966d5523d", "source_url": "https://huggingface.co/LGAI-EXAONE/K-EXAONE-2.0-750B-A37B", "org": "LGAI-EXAONE", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-07-31", "precision": "day", "url": "https://github.com/LG-AI-EXAONE/K-EXAONE-2.0", "status": "verified_primary", "note": "Official News entry; repository created July29 is not release." }, "params": { "total_b": 750, "active_b": 37, "moe": true, "url": "https://huggingface.co/LGAI-EXAONE/K-EXAONE-2.0-750B-A37B/raw/fec0c8dd25fdd20ffd09869d5af5277966d5523d/README.md", "active_url": "https://huggingface.co/LGAI-EXAONE/K-EXAONE-2.0-750B-A37B/raw/fec0c8dd25fdd20ffd09869d5af5277966d5523d/README.md", "scope": null }, "arch_label": "MoE · sliding-window attn", "context": { "native": 262144, "url": "https://huggingface.co/LGAI-EXAONE/K-EXAONE-2.0-750B-A37B/raw/fec0c8dd25fdd20ffd09869d5af5277966d5523d/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/LGAI-EXAONE/K-EXAONE-2.0-750B-A37B/blob/fec0c8dd25fdd20ffd09869d5af5277966d5523d/LICENSE" }, "quality": { "rank": 34, "aa": { "value": 19.6904943371003, "variant": "K-EXAONE 2.0 0803", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/k-exaone-2-0-0803", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/LGAI-EXAONE/K-EXAONE-2.0-750B-A37B/raw/fec0c8dd25fdd20ffd09869d5af5277966d5523d/config.json", "dtype": "BF16", "note": "BF16/FP16 KV estimate for full-attention layers only, batch=1. Sliding-window terms and fixed recurrent state are separate; excludes alignment, graphs, buffers, vision and MTP/drafters." }, "artifacts": {}, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_Q8_0", "GGUF_FP16", "GGUF_BF16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "longcat-2.0", "name": "LongCat-2.0", "hf_repo": "meituan-longcat/LongCat-2.0", "hf_revision": "834bf5ffe3047aa9f6cc7a64a9bc068b146b8274", "source_url": "https://huggingface.co/meituan-longcat/LongCat-2.0", "org": "meituan-longcat", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-07", "precision": "month", "url": "https://tech.meituan.com/2026/07/28/CatPaw-LongCat.html", "status": "verified_publisher_release_month", "note": "June30 announcement said will open-source; July28 official retrospective explicitly says open-sourced this month. Exact weight day unknown." }, "params": { "total_b": 1600, "active_b": 48, "moe": true, "url": "https://huggingface.co/meituan-longcat/LongCat-2.0/raw/834bf5ffe3047aa9f6cc7a64a9bc068b146b8274/README.md", "active_url": "https://huggingface.co/meituan-longcat/LongCat-2.0/raw/834bf5ffe3047aa9f6cc7a64a9bc068b146b8274/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 262144, "url": "https://huggingface.co/meituan-longcat/LongCat-2.0/raw/834bf5ffe3047aa9f6cc7a64a9bc068b146b8274/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/meituan-longcat/LongCat-2.0/blob/834bf5ffe3047aa9f6cc7a64a9bc068b146b8274/LICENSE" }, "quality": { "rank": 35, "aa": { "value": 19.1123256610623, "variant": "LongCat 2.0", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/longcat-2-0", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/meituan-longcat/LongCat-2.0/raw/834bf5ffe3047aa9f6cc7a64a9bc068b146b8274/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": {}, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_Q8_0", "GGUF_FP16", "GGUF_BF16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [ "Publisher reports 1.6T and also 135B n-gram embeddings; serialized tensor total differs. Preserve official scope." ] }, { "slug": "qwen3.6-35b-a3b", "name": "Qwen3.6-35B-A3B", "hf_repo": "Qwen/Qwen3.6-35B-A3B", "hf_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0", "source_url": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B", "org": "Qwen", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-04-16", "precision": "day", "url": "https://github.com/QwenLM/Qwen3.8/blob/main/README.md", "status": "verified_primary", "note": "Official dated release-news entry states public model availability." }, "params": { "total_b": 35, "active_b": 3, "moe": true, "url": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B/blob/995ad96eacd98c81ed38be0c5b274b04031597b0/README.md", "active_url": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B/blob/995ad96eacd98c81ed38be0c5b274b04031597b0/README.md", "scope": "Publisher nominal language-model size; HF total includes stored multimodal/MTP tensors as applicable." }, "arch_label": "MoE · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B/blob/995ad96eacd98c81ed38be0c5b274b04031597b0/config.json", "extended": 1010000 }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B/blob/main/LICENSE" }, "quality": { "rank": 36, "aa": { "value": 18.2289900189843, "variant": "Qwen3.6 35B A3B (Reasoning)", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/qwen3-6-35b-a3b", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": 20480, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B/blob/995ad96eacd98c81ed38be0c5b274b04031597b0/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 20419565568, "repo": "ggml-org/Qwen3.6-35B-A3B-GGUF", "revision": "baec3ebee244827cda0f4557eafa8b28f7545fa6", "url": "https://huggingface.co/ggml-org/Qwen3.6-35B-A3B-GGUF/tree/baec3ebee244827cda0f4557eafa8b28f7545fa6", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-Qwen3.6-35B-A3B-BF16.gguf", "bytes": 902822144, "url": "https://huggingface.co/ggml-org/Qwen3.6-35B-A3B-GGUF/blob/baec3ebee244827cda0f4557eafa8b28f7545fa6/mmproj-Qwen3.6-35B-A3B-BF16.gguf" }, { "path": "mmproj-Qwen3.6-35B-A3B-Q8_0.gguf", "bytes": 614194304, "url": "https://huggingface.co/ggml-org/Qwen3.6-35B-A3B-GGUF/blob/baec3ebee244827cda0f4557eafa8b28f7545fa6/mmproj-Qwen3.6-35B-A3B-Q8_0.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 24729131296, "repo": "AtomicChat/Qwen3.6-35B-A3B-GGUF", "revision": "9f7a1a1c3c8f3e2dd495304a964b92c012021250", "url": "https://huggingface.co/AtomicChat/Qwen3.6-35B-A3B-GGUF/tree/9f7a1a1c3c8f3e2dd495304a964b92c012021250", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 36903140320, "repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "revision": "a483e9e6cbd595906af30beda3187c2663a1118c", "url": "https://huggingface.co/unsloth/Qwen3.6-35B-A3B-GGUF/tree/a483e9e6cbd595906af30beda3187c2663a1118c", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 902822624, "url": "https://huggingface.co/unsloth/Qwen3.6-35B-A3B-GGUF/blob/a483e9e6cbd595906af30beda3187c2663a1118c/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 899283680, "url": "https://huggingface.co/unsloth/Qwen3.6-35B-A3B-GGUF/blob/a483e9e6cbd595906af30beda3187c2663a1118c/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1786305248, "url": "https://huggingface.co/unsloth/Qwen3.6-35B-A3B-GGUF/blob/a483e9e6cbd595906af30beda3187c2663a1118c/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 20402204271, "repo": "mlx-community/Qwen3.6-35B-A3B-4bit", "revision": "38740b847e4cb78f352aba30aa41c76e08e6eb46", "url": "https://huggingface.co/mlx-community/Qwen3.6-35B-A3B-4bit/tree/38740b847e4cb78f352aba30aa41c76e08e6eb46", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 4, "status": "verified_hf_file_sizes", "mixed": true, "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 37721403002, "repo": "mlx-community/Qwen3.6-35B-A3B-8bit", "revision": "e06a74e6236a60c8367e1a3214e83d8b61b637b0", "url": "https://huggingface.co/mlx-community/Qwen3.6-35B-A3B-8bit/tree/e06a74e6236a60c8367e1a3214e83d8b61b637b0", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 8, "status": "verified_hf_file_sizes", "mixed": true, "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-05-22", "title": "Possible issue: completion-only output may contain without a preceding due to chat template prefill", "url": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B/discussions/62" }, { "platform": "Hugging Face", "date": "2026-05-19", "title": "Qwen3.6 support for TRT-LLM?", "url": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B/discussions/60" } ], "notes": [] }, { "slug": "qwen3-5-122b-a10b", "name": "Qwen3.5-122B-A10B", "hf_repo": "Qwen/Qwen3.5-122B-A10B", "hf_revision": "dc4d348443bc740c68e2d77492492c11606384d5", "source_url": "https://huggingface.co/Qwen/Qwen3.5-122B-A10B", "org": "Qwen", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-02-24", "precision": "day", "url": "https://github.com/QwenLM/Qwen3.8", "status": "verified_primary", "note": null }, "params": { "total_b": 122, "active_b": 10, "moe": true, "url": "https://huggingface.co/Qwen/Qwen3.5-122B-A10B/raw/dc4d348443bc740c68e2d77492492c11606384d5/README.md", "active_url": "https://huggingface.co/Qwen/Qwen3.5-122B-A10B/raw/dc4d348443bc740c68e2d77492492c11606384d5/README.md", "scope": null }, "arch_label": "MoE · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/Qwen/Qwen3.5-122B-A10B/raw/dc4d348443bc740c68e2d77492492c11606384d5/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/Qwen/Qwen3.5-122B-A10B/blob/main/LICENSE" }, "quality": { "rank": 37, "aa": { "value": 17.7480154665397, "variant": "Qwen3.5 122B A10B (Non-reasoning)", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/qwen3-5-122b-a10b-non-reasoning", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": { "value": 1417.3957431609, "lower": 1413.1360765763, "upper": 1421.6554097454, "votes": 29732, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": 24576, "win_per_token": null, "win_tokens": null, "win_max": 0, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3.5-122B-A10B/raw/dc4d348443bc740c68e2d77492492c11606384d5/config.json", "dtype": "BF16", "note": "BF16/FP16 KV estimate for full-attention layers only, batch=1. Sliding-window terms and fixed recurrent state are separate; excludes alignment, graphs, buffers, vision and MTP/drafters." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 76536964608, "repo": "unsloth/Qwen3.5-122B-A10B-GGUF", "revision": "51eab4d59d53f573fb9206cb3ce613f1d0aa392b", "url": "https://huggingface.co/unsloth/Qwen3.5-122B-A10B-GGUF/tree/51eab4d59d53f573fb9206cb3ce613f1d0aa392b", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 3, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 912263904, "url": "https://huggingface.co/unsloth/Qwen3.5-122B-A10B-GGUF/resolve/51eab4d59d53f573fb9206cb3ce613f1d0aa392b/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 908724960, "url": "https://huggingface.co/unsloth/Qwen3.5-122B-A10B-GGUF/resolve/51eab4d59d53f573fb9206cb3ce613f1d0aa392b/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1805183712, "url": "https://huggingface.co/unsloth/Qwen3.5-122B-A10B-GGUF/resolve/51eab4d59d53f573fb9206cb3ce613f1d0aa392b/mmproj-F32.gguf" } ] }, "GGUF_Q5_K_M": { "bytes": 91519219200, "repo": "unsloth/Qwen3.5-122B-A10B-GGUF", "revision": "51eab4d59d53f573fb9206cb3ce613f1d0aa392b", "url": "https://huggingface.co/unsloth/Qwen3.5-122B-A10B-GGUF/tree/51eab4d59d53f573fb9206cb3ce613f1d0aa392b", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 3, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 912263904, "url": "https://huggingface.co/unsloth/Qwen3.5-122B-A10B-GGUF/resolve/51eab4d59d53f573fb9206cb3ce613f1d0aa392b/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 908724960, "url": "https://huggingface.co/unsloth/Qwen3.5-122B-A10B-GGUF/resolve/51eab4d59d53f573fb9206cb3ce613f1d0aa392b/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1805183712, "url": "https://huggingface.co/unsloth/Qwen3.5-122B-A10B-GGUF/resolve/51eab4d59d53f573fb9206cb3ce613f1d0aa392b/mmproj-F32.gguf" } ] }, "GGUF_Q8_0": { "bytes": 129871935104, "repo": "unsloth/Qwen3.5-122B-A10B-GGUF", "revision": "51eab4d59d53f573fb9206cb3ce613f1d0aa392b", "url": "https://huggingface.co/unsloth/Qwen3.5-122B-A10B-GGUF/tree/51eab4d59d53f573fb9206cb3ce613f1d0aa392b", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 4, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 912263904, "url": "https://huggingface.co/unsloth/Qwen3.5-122B-A10B-GGUF/resolve/51eab4d59d53f573fb9206cb3ce613f1d0aa392b/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 908724960, "url": "https://huggingface.co/unsloth/Qwen3.5-122B-A10B-GGUF/resolve/51eab4d59d53f573fb9206cb3ce613f1d0aa392b/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1805183712, "url": "https://huggingface.co/unsloth/Qwen3.5-122B-A10B-GGUF/resolve/51eab4d59d53f573fb9206cb3ce613f1d0aa392b/mmproj-F32.gguf" } ] }, "GGUF_BF16": { "bytes": 244314011488, "repo": "unsloth/Qwen3.5-122B-A10B-GGUF", "revision": "51eab4d59d53f573fb9206cb3ce613f1d0aa392b", "url": "https://huggingface.co/unsloth/Qwen3.5-122B-A10B-GGUF/tree/51eab4d59d53f573fb9206cb3ce613f1d0aa392b", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 5, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 912263904, "url": "https://huggingface.co/unsloth/Qwen3.5-122B-A10B-GGUF/resolve/51eab4d59d53f573fb9206cb3ce613f1d0aa392b/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 908724960, "url": "https://huggingface.co/unsloth/Qwen3.5-122B-A10B-GGUF/resolve/51eab4d59d53f573fb9206cb3ce613f1d0aa392b/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1805183712, "url": "https://huggingface.co/unsloth/Qwen3.5-122B-A10B-GGUF/resolve/51eab4d59d53f573fb9206cb3ce613f1d0aa392b/mmproj-F32.gguf" } ] }, "MLX_4bit": { "bytes": 69593636939, "repo": "mlx-community/Qwen3.5-122B-A10B-4bit", "revision": "e9c67b08899964be5fdd069bb1b4bc8907fe68f5", "url": "https://huggingface.co/mlx-community/Qwen3.5-122B-A10B-4bit/tree/e9c67b08899964be5fdd069bb1b4bc8907fe68f5", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 14, "status": "measured_file_manifest" }, "MLX_8bit": { "bytes": 130648358613, "repo": "mlx-community/Qwen3.5-122B-A10B-8bit", "revision": "2a6260e0a7c3e80a1acc401999372aadeb870cff", "url": "https://huggingface.co/mlx-community/Qwen3.5-122B-A10B-8bit/tree/2a6260e0a7c3e80a1acc401999372aadeb870cff", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 25, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [], "notes": [] }, { "slug": "muse-glimmer-30b", "name": "Muse Glimmer 30B", "hf_repo": "meta-models/Muse-Glimmer-30B", "hf_revision": "a4e59da52a7bc87ae7251dd5545c0dd437c44b68", "source_url": "https://huggingface.co/meta-models/Muse-Glimmer-30B", "org": "meta-models", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08-10", "precision": "day", "url": "https://research.meta.ai/blog/introducing-muse-glimmer-open-agentic-model", "status": "verified_primary", "note": "Publisher announcement." }, "params": { "total_b": 29.6, "active_b": 29.6, "moe": false, "url": "https://huggingface.co/meta-models/Muse-Glimmer-30B/blob/a4e59da52a7bc87ae7251dd5545c0dd437c44b68/README.md", "active_url": null, "scope": "Publisher model card says approximately 29.6B including vision encoder; API tensor total differs." }, "arch_label": "Dense · sliding-window attn", "context": { "native": 131072, "url": "https://huggingface.co/meta-models/Muse-Glimmer-30B/blob/a4e59da52a7bc87ae7251dd5545c0dd437c44b68/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/meta-models/Muse-Glimmer-30B/blob/a4e59da52a7bc87ae7251dd5545c0dd437c44b68/LICENSE" }, "quality": { "rank": 38, "aa": { "value": 17.4754176089465, "variant": "Muse Glimmer (High)", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/muse-glimmer", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1388.9089387323, "lower": 1379.2647581669, "upper": 1398.5531192977, "votes": 3897, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": 13312, "win_per_token": 39936, "win_tokens": 2048, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/meta-models/Muse-Glimmer-30B/blob/a4e59da52a7bc87ae7251dd5545c0dd437c44b68/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 16756683904, "repo": "meta-models/Muse-Glimmer-30B-GGUF", "revision": "70bf1b61ac09f91b24d39038091b41c582bc5d7a", "url": "https://huggingface.co/meta-models/Muse-Glimmer-30B-GGUF/tree/70bf1b61ac09f91b24d39038091b41c582bc5d7a", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-Muse-Glimmer-30B-Q4_K_M.gguf", "bytes": 1400328928, "url": "https://huggingface.co/meta-models/Muse-Glimmer-30B-GGUF/blob/70bf1b61ac09f91b24d39038091b41c582bc5d7a/mmproj-Muse-Glimmer-30B-Q4_K_M.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 20105571360, "repo": "bartowski/Muse-Glimmer-30B-GGUF", "revision": "01e1b7c30e19e1823ee9a73dec51d12f59ec4857", "url": "https://huggingface.co/bartowski/Muse-Glimmer-30B-GGUF/tree/01e1b7c30e19e1823ee9a73dec51d12f59ec4857", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-Muse-Glimmer-30B-bf16.gguf", "bytes": 3849173920, "url": "https://huggingface.co/bartowski/Muse-Glimmer-30B-GGUF/blob/01e1b7c30e19e1823ee9a73dec51d12f59ec4857/mmproj-Muse-Glimmer-30B-bf16.gguf" }, { "path": "mmproj-Muse-Glimmer-30B-f16.gguf", "bytes": 3847367584, "url": "https://huggingface.co/bartowski/Muse-Glimmer-30B-GGUF/blob/01e1b7c30e19e1823ee9a73dec51d12f59ec4857/mmproj-Muse-Glimmer-30B-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 29612958496, "repo": "bartowski/Muse-Glimmer-30B-GGUF", "revision": "01e1b7c30e19e1823ee9a73dec51d12f59ec4857", "url": "https://huggingface.co/bartowski/Muse-Glimmer-30B-GGUF/tree/01e1b7c30e19e1823ee9a73dec51d12f59ec4857", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-Muse-Glimmer-30B-bf16.gguf", "bytes": 3849173920, "url": "https://huggingface.co/bartowski/Muse-Glimmer-30B-GGUF/blob/01e1b7c30e19e1823ee9a73dec51d12f59ec4857/mmproj-Muse-Glimmer-30B-bf16.gguf" }, { "path": "mmproj-Muse-Glimmer-30B-f16.gguf", "bytes": 3847367584, "url": "https://huggingface.co/bartowski/Muse-Glimmer-30B-GGUF/blob/01e1b7c30e19e1823ee9a73dec51d12f59ec4857/mmproj-Muse-Glimmer-30B-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 19414804113, "repo": "mlx-community/Muse-Glimmer-30B-4bit", "revision": "3e7677d7a40d348a3daba263a2b1c0aa41910710", "url": "https://huggingface.co/mlx-community/Muse-Glimmer-30B-4bit/tree/3e7677d7a40d348a3daba263a2b1c0aa41910710", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 4, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 33376100422, "repo": "mlx-community/Muse-Glimmer-30B-8bit", "revision": "679c45e1b331a6514a6e38076a353cc5fed21bf6", "url": "https://huggingface.co/mlx-community/Muse-Glimmer-30B-8bit/tree/679c45e1b331a6514a6e38076a353cc5fed21bf6", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 153, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-09-14", "title": "Training LoRAs on Muse-Glimmer for llama.cpp: 5 things that cost us a week (label format, reasoning_strength, peg-native 500s, convert_lora_to_gguf crash, recipe)", "url": "https://huggingface.co/meta-models/Muse-Glimmer-30B/discussions/63" }, { "platform": "Hugging Face", "date": "2026-08-13", "title": "Tutorial on how to run with llama.cpp", "url": "https://huggingface.co/meta-models/Muse-Glimmer-30B/discussions/53" } ], "notes": [] }, { "slug": "gemma-4-26b-a4b", "name": "Gemma 4 26B A4B (it)", "hf_repo": "google/gemma-4-26B-A4B-it", "hf_revision": "4d7ae4984b7db7de8f8457170b3f1a419ee76d52", "source_url": "https://huggingface.co/google/gemma-4-26B-A4B-it", "org": "google", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-04-02", "precision": "day", "url": "https://blog.google/innovation-and-ai/technology/developers-tools/gemma-4/", "status": "verified_primary", "note": "Public launch announcement; distinguish earlier repository creation." }, "params": { "total_b": 25.2, "active_b": 3.8, "moe": true, "url": "https://huggingface.co/google/gemma-4-26B-A4B-it/blob/4d7ae4984b7db7de8f8457170b3f1a419ee76d52/README.md", "active_url": "https://huggingface.co/google/gemma-4-26B-A4B-it/blob/4d7ae4984b7db7de8f8457170b3f1a419ee76d52/README.md", "scope": "Publisher decoder count plus approximately 0.55B vision encoder; nominal model name rounds to 26B." }, "arch_label": "MoE · sliding-window attn", "context": { "native": 262144, "url": "https://huggingface.co/google/gemma-4-26B-A4B-it/blob/4d7ae4984b7db7de8f8457170b3f1a419ee76d52/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://ai.google.dev/gemma/docs/gemma_4_license" }, "quality": { "rank": 39, "aa": { "value": 16.671930040086, "variant": "Gemma 4 26B A4B (Reasoning)", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/gemma-4-26b-a4b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": { "value": 1434.1159973589, "lower": 1426.7003934669, "upper": 1441.531601251, "votes": 6094, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/google/gemma-4-26B-A4B-it/blob/4d7ae4984b7db7de8f8457170b3f1a419ee76d52/config.json", "dtype": "FP16", "note": "Architecture needs backend-specific treatment: shared K/V or cross-layer cache, looped layers, MoVA, MLA/KDA or diffusion. No naive all-layers GQA estimate is supplied." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 16796017248, "repo": "lmstudio-community/gemma-4-26B-A4B-it-GGUF", "revision": "f6e6747823b2912661935db7e0009287c4838073", "url": "https://huggingface.co/lmstudio-community/gemma-4-26B-A4B-it-GGUF/tree/f6e6747823b2912661935db7e0009287c4838073", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-gemma-4-26B-A4B-it-BF16.gguf", "bytes": 1194827744, "url": "https://huggingface.co/lmstudio-community/gemma-4-26B-A4B-it-GGUF/blob/f6e6747823b2912661935db7e0009287c4838073/mmproj-gemma-4-26B-A4B-it-BF16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 26859861728, "repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "revision": "c099eb48e663fd284577b04978a94ffccb261841", "url": "https://huggingface.co/unsloth/gemma-4-26B-A4B-it-GGUF/tree/c099eb48e663fd284577b04978a94ffccb261841", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 1194828256, "url": "https://huggingface.co/unsloth/gemma-4-26B-A4B-it-GGUF/blob/c099eb48e663fd284577b04978a94ffccb261841/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 1193058784, "url": "https://huggingface.co/unsloth/gemma-4-26B-A4B-it-GGUF/blob/c099eb48e663fd284577b04978a94ffccb261841/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 2291200480, "url": "https://huggingface.co/unsloth/gemma-4-26B-A4B-it-GGUF/blob/c099eb48e663fd284577b04978a94ffccb261841/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 15341205776, "repo": "mlx-community/gemma-4-26B-A4B-it-4bit", "revision": "0d77464eeb233a2da68ebf9d7dc4edaac7db956d", "url": "https://huggingface.co/mlx-community/gemma-4-26B-A4B-it-4bit/tree/0d77464eeb233a2da68ebf9d7dc4edaac7db956d", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 3, "status": "verified_hf_file_sizes", "mixed": true, "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 27953641429, "repo": "mlx-community/gemma-4-26B-A4B-it-8bit", "revision": "33c6d23798a0af159529890f79329206dbfbd73c", "url": "https://huggingface.co/mlx-community/gemma-4-26B-A4B-it-8bit/tree/33c6d23798a0af159529890f79329206dbfbd73c", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 6, "status": "verified_hf_file_sizes", "mixed": true, "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_Q5_K_M", "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-09-23", "title": "Reliable tool calling for Markdown vaults and LLM wikis w/ Gemma", "url": "https://huggingface.co/google/gemma-4-26B-A4B-it/discussions/65" }, { "platform": "Hugging Face", "date": "2026-06-10", "title": "Chat template may re-inject prior-turn reasoning during multi-turn tool use → repetition loops", "url": "https://huggingface.co/google/gemma-4-26B-A4B-it/discussions/48" } ], "notes": [] }, { "slug": "ring-2.6-1t", "name": "Ring-2.6-1T", "hf_repo": "inclusionAI/Ring-2.6-1T", "hf_revision": "2314abdf9e380b7fa7b4b79369ceb5f69cc46b32", "source_url": "https://huggingface.co/inclusionAI/Ring-2.6-1T", "org": "inclusionAI", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-05-15", "precision": "day", "url": "https://www.ant-ling.com/zh/", "status": "verified_primary", "note": "Publisher dated news explicitly labels open-source release." }, "params": { "total_b": 1000, "active_b": 63, "moe": true, "url": "https://huggingface.co/inclusionAI/Ring-2.6-1T/raw/2314abdf9e380b7fa7b4b79369ceb5f69cc46b32/README.md", "active_url": "https://developer.ant-ling.com/en/docs/models/ring/", "scope": null }, "arch_label": "MoE", "context": { "native": 131072, "url": "https://huggingface.co/inclusionAI/Ring-2.6-1T/raw/2314abdf9e380b7fa7b4b79369ceb5f69cc46b32/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/inclusionAI/Ring-2.6-1T/blob/2314abdf9e380b7fa7b4b79369ceb5f69cc46b32/LICENSE" }, "quality": { "rank": 40, "aa": { "value": 16.6172363768618, "variant": "Ring-2.6-1T", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/ring-2-6-1t", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/inclusionAI/Ring-2.6-1T/raw/2314abdf9e380b7fa7b4b79369ceb5f69cc46b32/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": {}, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_Q8_0", "GGUF_FP16", "GGUF_BF16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [ "Publisher calls this a trillion-parameter model; active parameters not verified." ] }, { "slug": "k2-horizon-3.7b", "name": "K2-Horizon-3.7B", "hf_repo": "IFM/K2-Horizon-3.7B", "hf_revision": "9d280ff742235796925c9fe3cf243dc1beea5c08", "source_url": "https://huggingface.co/IFM/K2-Horizon-3.7B", "org": "IFM", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-09-03", "precision": "day", "url": "https://mbzuai.ac.ae/news-events/news/mbzuais-institute-foundation-models-launches-k2-horizon-the-worlds-largest-fully", "status": "verified_primary", "note": "Publisher launch announces six models available immediately, including this size." }, "params": { "total_b": null, "active_b": null, "moe": false, "url": null, "active_url": null, "scope": "3.7B is core/non-embedding class; not total resident weights." }, "arch_label": "Dense", "context": { "native": 524288, "url": "https://huggingface.co/IFM/K2-Horizon-3.7B/blob/9d280ff742235796925c9fe3cf243dc1beea5c08/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/IFM/K2-Horizon-3.7B/blob/9d280ff742235796925c9fe3cf243dc1beea5c08/LICENSE" }, "quality": { "rank": 41, "aa": { "value": 15.6103951330976, "variant": "K2 Horizon 3.7B", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/k2-horizon-3-7b", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": 147456, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/IFM/K2-Horizon-3.7B/blob/9d280ff742235796925c9fe3cf243dc1beea5c08/config.json", "dtype": "FP16", "note": "Ordinary GQA cache estimate; actual allocation and padding vary by backend." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 3156598144, "repo": "IFM/K2-Horizon-3.7B-GGUF", "revision": "78a0635e8f19b880e468a2fc82923ee6c8a469e4", "url": "https://huggingface.co/IFM/K2-Horizon-3.7B-GGUF/tree/78a0635e8f19b880e468a2fc82923ee6c8a469e4", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 3643776384, "repo": "IFM/K2-Horizon-3.7B-GGUF", "revision": "78a0635e8f19b880e468a2fc82923ee6c8a469e4", "url": "https://huggingface.co/IFM/K2-Horizon-3.7B-GGUF/tree/78a0635e8f19b880e468a2fc82923ee6c8a469e4", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 5386404224, "repo": "IFM/K2-Horizon-3.7B-GGUF", "revision": "78a0635e8f19b880e468a2fc82923ee6c8a469e4", "url": "https://huggingface.co/IFM/K2-Horizon-3.7B-GGUF/tree/78a0635e8f19b880e468a2fc82923ee6c8a469e4", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 10128352672, "repo": "mradermacher/K2-Horizon-3.7B-GGUF", "revision": "3b4d8f2c125415d6cd68c309bf02bc278fac80c9", "url": "https://huggingface.co/mradermacher/K2-Horizon-3.7B-GGUF/tree/3b4d8f2c125415d6cd68c309bf02bc278fac80c9", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 4136036684, "repo": "mlx-community/K2-Horizon-3.7B-4bit", "revision": "3b453d026bd4f433c0bc9a532706f87488269e4d", "url": "https://huggingface.co/mlx-community/K2-Horizon-3.7B-4bit/tree/3b453d026bd4f433c0bc9a532706f87488269e4d", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "mixed": true, "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 5374667178, "repo": "mlx-community/K2-Horizon-3.7B-8bit", "revision": "6933978e66e52f5ed0d0cb768969d94d44aceb95", "url": "https://huggingface.co/mlx-community/K2-Horizon-3.7B-8bit/tree/6933978e66e52f5ed0d0cb768969d94d44aceb95", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 2, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-09-04", "title": "Can we have GGUF version", "url": "https://huggingface.co/IFM/K2-Horizon-3.7B/discussions/3" }, { "platform": "Hugging Face", "date": "2026-10-01", "title": "Data mixture", "url": "https://huggingface.co/IFM/K2-Horizon-3.7B/discussions/5" } ], "notes": [] }, { "slug": "gemma-4-31b", "name": "Gemma 4 31B (it)", "hf_repo": "google/gemma-4-31B-it", "hf_revision": "842da3794eaa0b77d5f08bae87a17459d91ff475", "source_url": "https://huggingface.co/google/gemma-4-31B-it", "org": "google", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-04-02", "precision": "day", "url": "https://blog.google/innovation-and-ai/technology/developers-tools/gemma-4/", "status": "verified_primary", "note": "Public launch announcement; distinguish earlier repository creation." }, "params": { "total_b": 30.7, "active_b": 30.7, "moe": false, "url": "https://huggingface.co/google/gemma-4-31B-it/blob/842da3794eaa0b77d5f08bae87a17459d91ff475/README.md", "active_url": null, "scope": "Publisher decoder count; additional approximately 0.55B vision encoder; whole-checkpoint HF total separate." }, "arch_label": "Dense · sliding-window attn", "context": { "native": 262144, "url": "https://huggingface.co/google/gemma-4-31B-it/blob/842da3794eaa0b77d5f08bae87a17459d91ff475/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://ai.google.dev/gemma/docs/gemma_4_license" }, "quality": { "rank": 42, "aa": { "value": 14.6689500745828, "variant": "Gemma 4 31B (Reasoning)", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/gemma-4-31b", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1442.863540914, "lower": 1435.471128224, "upper": 1450.2559536041, "votes": 6147, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/google/gemma-4-31B-it/blob/842da3794eaa0b77d5f08bae87a17459d91ff475/config.json", "dtype": "FP16", "note": "Architecture needs backend-specific treatment: shared K/V or cross-layer cache, looped layers, MoVA, MLA/KDA or diffusion. No naive all-layers GQA estimate is supplied." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 18323733440, "repo": "unsloth/gemma-4-31B-it-GGUF", "revision": "c1ac76e99d5513b141e8adde7288b85c3f9c32ec", "url": "https://huggingface.co/unsloth/gemma-4-31B-it-GGUF/tree/c1ac76e99d5513b141e8adde7288b85c3f9c32ec", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 1200726496, "url": "https://huggingface.co/unsloth/gemma-4-31B-it-GGUF/blob/c1ac76e99d5513b141e8adde7288b85c3f9c32ec/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 1198957024, "url": "https://huggingface.co/unsloth/gemma-4-31B-it-GGUF/blob/c1ac76e99d5513b141e8adde7288b85c3f9c32ec/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 2302996960, "url": "https://huggingface.co/unsloth/gemma-4-31B-it-GGUF/blob/c1ac76e99d5513b141e8adde7288b85c3f9c32ec/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 21658401728, "repo": "unsloth/gemma-4-31B-it-GGUF", "revision": "c1ac76e99d5513b141e8adde7288b85c3f9c32ec", "url": "https://huggingface.co/unsloth/gemma-4-31B-it-GGUF/tree/c1ac76e99d5513b141e8adde7288b85c3f9c32ec", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 1200726496, "url": "https://huggingface.co/unsloth/gemma-4-31B-it-GGUF/blob/c1ac76e99d5513b141e8adde7288b85c3f9c32ec/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 1198957024, "url": "https://huggingface.co/unsloth/gemma-4-31B-it-GGUF/blob/c1ac76e99d5513b141e8adde7288b85c3f9c32ec/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 2302996960, "url": "https://huggingface.co/unsloth/gemma-4-31B-it-GGUF/blob/c1ac76e99d5513b141e8adde7288b85c3f9c32ec/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 32635677632, "repo": "unsloth/gemma-4-31B-it-GGUF", "revision": "c1ac76e99d5513b141e8adde7288b85c3f9c32ec", "url": "https://huggingface.co/unsloth/gemma-4-31B-it-GGUF/tree/c1ac76e99d5513b141e8adde7288b85c3f9c32ec", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 1200726496, "url": "https://huggingface.co/unsloth/gemma-4-31B-it-GGUF/blob/c1ac76e99d5513b141e8adde7288b85c3f9c32ec/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 1198957024, "url": "https://huggingface.co/unsloth/gemma-4-31B-it-GGUF/blob/c1ac76e99d5513b141e8adde7288b85c3f9c32ec/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 2302996960, "url": "https://huggingface.co/unsloth/gemma-4-31B-it-GGUF/blob/c1ac76e99d5513b141e8adde7288b85c3f9c32ec/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 18412016676, "repo": "mlx-community/gemma-4-31B-it-4bit", "revision": "696d436c404745a59f30e4939a658162b0a9e57f", "url": "https://huggingface.co/mlx-community/gemma-4-31B-it-4bit/tree/696d436c404745a59f30e4939a658162b0a9e57f", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 4, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 33763120415, "repo": "mlx-community/gemma-4-31B-it-8bit", "revision": "f5f3dc92ab4af76724c36c21eb6bedadb3a851be", "url": "https://huggingface.co/mlx-community/gemma-4-31B-it-8bit/tree/f5f3dc92ab4af76724c36c21eb6bedadb3a851be", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 7, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-07-30", "title": "Same tool call infinite loop", "url": "https://huggingface.co/google/gemma-4-31B-it/discussions/139" }, { "platform": "Hugging Face", "date": "2026-07-23", "title": "When Gemma doesn't have the right tool schema it hallucinates it, create outputs of the fake-calls and use them as memory (WIP)", "url": "https://huggingface.co/google/gemma-4-31B-it/discussions/136" } ], "notes": [] }, { "slug": "mistral-medium-3-5", "name": "Mistral Medium 3.5 (128B dense)", "hf_repo": "mistralai/Mistral-Medium-3.5-128B", "hf_revision": "22b2b868a15677cfa6061277ed2f653d1349a9ab", "source_url": "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B", "org": "mistralai", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-04-28", "precision": "day", "url": "https://docs.mistral.ai/models/mistral-medium-3-5-26-04", "status": "verified_primary", "note": "Official docs and changelog say Apr28; current launch blog page displays May22. Preserve source discrepancy." }, "params": { "total_b": 128, "active_b": 128, "moe": false, "url": "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B/raw/22b2b868a15677cfa6061277ed2f653d1349a9ab/README.md", "active_url": "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B/raw/22b2b868a15677cfa6061277ed2f653d1349a9ab/README.md", "scope": null }, "arch_label": "Dense", "context": { "native": 262144, "url": "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B/raw/22b2b868a15677cfa6061277ed2f653d1349a9ab/config.json", "extended": null }, "license": { "value": "other", "url": "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B/blob/22b2b868a15677cfa6061277ed2f653d1349a9ab/LICENSE" }, "quality": { "rank": 43, "aa": { "value": 14.1889227802691, "variant": "Mistral Medium 3.5", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/mistral-medium-3-5", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1420.966198361, "lower": 1414.5900912118, "upper": 1427.3423055101, "votes": 11778, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": 360448, "win_per_token": null, "win_tokens": null, "win_max": 0, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B/raw/22b2b868a15677cfa6061277ed2f653d1349a9ab/config.json", "dtype": "BF16", "note": "BF16/FP16 KV estimate for full-attention layers only, batch=1. Sliding-window terms and fixed recurrent state are separate; excludes alignment, graphs, buffers, vision and MTP/drafters." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 74897139136, "repo": "unsloth/Mistral-Medium-3.5-128B-GGUF", "revision": "c8f5b1477e1b22cd2d819157d450f001f7047298", "url": "https://huggingface.co/unsloth/Mistral-Medium-3.5-128B-GGUF/tree/c8f5b1477e1b22cd2d819157d450f001f7047298", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 3, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 5380954560, "url": "https://huggingface.co/unsloth/Mistral-Medium-3.5-128B-GGUF/resolve/c8f5b1477e1b22cd2d819157d450f001f7047298/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 5356846528, "url": "https://huggingface.co/unsloth/Mistral-Medium-3.5-128B-GGUF/resolve/c8f5b1477e1b22cd2d819157d450f001f7047298/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 10712963520, "url": "https://huggingface.co/unsloth/Mistral-Medium-3.5-128B-GGUF/resolve/c8f5b1477e1b22cd2d819157d450f001f7047298/mmproj-F32.gguf" } ] }, "GGUF_Q5_K_M": { "bytes": 88316814752, "repo": "unsloth/Mistral-Medium-3.5-128B-GGUF", "revision": "c8f5b1477e1b22cd2d819157d450f001f7047298", "url": "https://huggingface.co/unsloth/Mistral-Medium-3.5-128B-GGUF/tree/c8f5b1477e1b22cd2d819157d450f001f7047298", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 3, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 5380954560, "url": "https://huggingface.co/unsloth/Mistral-Medium-3.5-128B-GGUF/resolve/c8f5b1477e1b22cd2d819157d450f001f7047298/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 5356846528, "url": "https://huggingface.co/unsloth/Mistral-Medium-3.5-128B-GGUF/resolve/c8f5b1477e1b22cd2d819157d450f001f7047298/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 10712963520, "url": "https://huggingface.co/unsloth/Mistral-Medium-3.5-128B-GGUF/resolve/c8f5b1477e1b22cd2d819157d450f001f7047298/mmproj-F32.gguf" } ] }, "GGUF_Q8_0": { "bytes": 132854425088, "repo": "unsloth/Mistral-Medium-3.5-128B-GGUF", "revision": "c8f5b1477e1b22cd2d819157d450f001f7047298", "url": "https://huggingface.co/unsloth/Mistral-Medium-3.5-128B-GGUF/tree/c8f5b1477e1b22cd2d819157d450f001f7047298", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 4, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 5380954560, "url": "https://huggingface.co/unsloth/Mistral-Medium-3.5-128B-GGUF/resolve/c8f5b1477e1b22cd2d819157d450f001f7047298/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 5356846528, "url": "https://huggingface.co/unsloth/Mistral-Medium-3.5-128B-GGUF/resolve/c8f5b1477e1b22cd2d819157d450f001f7047298/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 10712963520, "url": "https://huggingface.co/unsloth/Mistral-Medium-3.5-128B-GGUF/resolve/c8f5b1477e1b22cd2d819157d450f001f7047298/mmproj-F32.gguf" } ] }, "GGUF_BF16": { "bytes": 250064250304, "repo": "unsloth/Mistral-Medium-3.5-128B-GGUF", "revision": "c8f5b1477e1b22cd2d819157d450f001f7047298", "url": "https://huggingface.co/unsloth/Mistral-Medium-3.5-128B-GGUF/tree/c8f5b1477e1b22cd2d819157d450f001f7047298", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 6, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 5380954560, "url": "https://huggingface.co/unsloth/Mistral-Medium-3.5-128B-GGUF/resolve/c8f5b1477e1b22cd2d819157d450f001f7047298/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 5356846528, "url": "https://huggingface.co/unsloth/Mistral-Medium-3.5-128B-GGUF/resolve/c8f5b1477e1b22cd2d819157d450f001f7047298/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 10712963520, "url": "https://huggingface.co/unsloth/Mistral-Medium-3.5-128B-GGUF/resolve/c8f5b1477e1b22cd2d819157d450f001f7047298/mmproj-F32.gguf" } ] }, "MLX_4bit": { "bytes": 78002274610, "repo": "mlx-community/Mistral-Medium-3.5-128B-4bit", "revision": "5a7b6b72ca9eb2324a90bdb5e6079789dae68348", "url": "https://huggingface.co/mlx-community/Mistral-Medium-3.5-128B-4bit/tree/5a7b6b72ca9eb2324a90bdb5e6079789dae68348", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 15, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_FP16", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "gemma-4-12b", "name": "Gemma 4 12B (it)", "hf_repo": "google/gemma-4-12B-it", "hf_revision": "707f0a3b8a3c7ad586ed01e27eafbad8a27dd0f7", "source_url": "https://huggingface.co/google/gemma-4-12B-it", "org": "google", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-06-03", "precision": "day", "url": "https://blog.google/innovation-and-ai/technology/developers-tools/introducing-gemma-4-12b/", "status": "verified_primary", "note": "Publisher announcement." }, "params": { "total_b": 11.95, "active_b": 11.95, "moe": false, "url": "https://huggingface.co/google/gemma-4-12B-it/blob/707f0a3b8a3c7ad586ed01e27eafbad8a27dd0f7/README.md", "active_url": null, "scope": "Dense unified multimodal model." }, "arch_label": "Dense · sliding-window attn", "context": { "native": 262144, "url": "https://huggingface.co/google/gemma-4-12B-it/blob/707f0a3b8a3c7ad586ed01e27eafbad8a27dd0f7/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://ai.google.dev/gemma/docs/gemma_4_license" }, "quality": { "rank": 44, "aa": { "value": 14.1847516647736, "variant": "Gemma 4 12B (Reasoning)", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/gemma-4-12b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/google/gemma-4-12B-it/blob/707f0a3b8a3c7ad586ed01e27eafbad8a27dd0f7/config.json", "dtype": "FP16", "note": "Architecture needs backend-specific treatment: shared K/V or cross-layer cache, looped layers, MoVA, MLA/KDA or diffusion. No naive all-layers GQA estimate is supplied." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 7662533088, "repo": "bartowski/gemma-4-12B-it-GGUF", "revision": "2ae7d41be21ca62de00a2d320ee9cec50daa3aa6", "url": "https://huggingface.co/bartowski/gemma-4-12B-it-GGUF/tree/2ae7d41be21ca62de00a2d320ee9cec50daa3aa6", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-gemma-4-12B-it-bf16.gguf", "bytes": 175115712, "url": "https://huggingface.co/bartowski/gemma-4-12B-it-GGUF/blob/2ae7d41be21ca62de00a2d320ee9cec50daa3aa6/mmproj-gemma-4-12B-it-bf16.gguf" }, { "path": "mmproj-gemma-4-12B-it-f16.gguf", "bytes": 122031552, "url": "https://huggingface.co/bartowski/gemma-4-12B-it-GGUF/blob/2ae7d41be21ca62de00a2d320ee9cec50daa3aa6/mmproj-gemma-4-12B-it-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 8774965728, "repo": "bartowski/gemma-4-12B-it-GGUF", "revision": "2ae7d41be21ca62de00a2d320ee9cec50daa3aa6", "url": "https://huggingface.co/bartowski/gemma-4-12B-it-GGUF/tree/2ae7d41be21ca62de00a2d320ee9cec50daa3aa6", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-gemma-4-12B-it-bf16.gguf", "bytes": 175115712, "url": "https://huggingface.co/bartowski/gemma-4-12B-it-GGUF/blob/2ae7d41be21ca62de00a2d320ee9cec50daa3aa6/mmproj-gemma-4-12B-it-bf16.gguf" }, { "path": "mmproj-gemma-4-12B-it-f16.gguf", "bytes": 122031552, "url": "https://huggingface.co/bartowski/gemma-4-12B-it-GGUF/blob/2ae7d41be21ca62de00a2d320ee9cec50daa3aa6/mmproj-gemma-4-12B-it-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 12669647328, "repo": "bartowski/gemma-4-12B-it-GGUF", "revision": "2ae7d41be21ca62de00a2d320ee9cec50daa3aa6", "url": "https://huggingface.co/bartowski/gemma-4-12B-it-GGUF/tree/2ae7d41be21ca62de00a2d320ee9cec50daa3aa6", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-gemma-4-12B-it-bf16.gguf", "bytes": 175115712, "url": "https://huggingface.co/bartowski/gemma-4-12B-it-GGUF/blob/2ae7d41be21ca62de00a2d320ee9cec50daa3aa6/mmproj-gemma-4-12B-it-bf16.gguf" }, { "path": "mmproj-gemma-4-12B-it-f16.gguf", "bytes": 122031552, "url": "https://huggingface.co/bartowski/gemma-4-12B-it-GGUF/blob/2ae7d41be21ca62de00a2d320ee9cec50daa3aa6/mmproj-gemma-4-12B-it-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 6741039511, "repo": "mlx-community/gemma-4-12B-it-4bit", "revision": "73bcf09092aa277861d5a191b989b666f7f32e8f", "url": "https://huggingface.co/mlx-community/gemma-4-12B-it-4bit/tree/73bcf09092aa277861d5a191b989b666f7f32e8f", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 2, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 12716202713, "repo": "mlx-community/gemma-4-12B-it-8bit", "revision": "200bb6db075e137a4deb08838865ac4ddb86292e", "url": "https://huggingface.co/mlx-community/gemma-4-12B-it-8bit/tree/200bb6db075e137a4deb08838865ac4ddb86292e", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 3, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-07-30", "title": "Bug Report: MAJOR issues with parallel generation and left-padding (different response from normal generatiln)", "url": "https://huggingface.co/google/gemma-4-12B-it/discussions/50" }, { "platform": "Hugging Face", "date": "2026-06-10", "title": "Chat template may re-inject prior-turn reasoning during multi-turn tool use → repetition loops", "url": "https://huggingface.co/google/gemma-4-12B-it/discussions/38" }, { "platform": "Reddit", "date": "2026-10-16", "title": null, "url": "https://www.reddit.com/r/LocalLLaMA/comments/1x7g8cc/why_do_we_have_models_under_15b_so_rarely_anymore/" }, { "platform": "Reddit", "date": "2026-06-03", "title": null, "url": "https://www.reddit.com/r/LocalLLM/comments/1tvx2h7/google_introduces_gemma_4_12b_a_unified/" } ], "notes": [] }, { "slug": "qwen3.5-9b", "name": "Qwen3.5-9B", "hf_repo": "Qwen/Qwen3.5-9B", "hf_revision": "c202236235762e1c871ad0ccb60c8ee5ba337b9a", "source_url": "https://huggingface.co/Qwen/Qwen3.5-9B", "org": "Qwen", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-03-02", "precision": "day", "url": "https://github.com/QwenLM/Qwen3.8/blob/main/README.md", "status": "verified_primary", "note": "Official dated release-news entry states public model availability." }, "params": { "total_b": 9, "active_b": 9, "moe": false, "url": "https://huggingface.co/Qwen/Qwen3.5-9B/blob/c202236235762e1c871ad0ccb60c8ee5ba337b9a/README.md", "active_url": null, "scope": "Publisher nominal language-model size; HF total includes stored multimodal/MTP tensors as applicable." }, "arch_label": "Dense · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/Qwen/Qwen3.5-9B/blob/c202236235762e1c871ad0ccb60c8ee5ba337b9a/config.json", "extended": 1010000 }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/Qwen/Qwen3.5-9B/blob/main/LICENSE" }, "quality": { "rank": 45, "aa": { "value": 13.2659955355465, "variant": "Qwen3.5 9B (Non-reasoning)", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/qwen3-5-9b-non-reasoning", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": 32768, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3.5-9B/blob/c202236235762e1c871ad0ccb60c8ee5ba337b9a/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 5680522464, "repo": "unsloth/Qwen3.5-9B-GGUF", "revision": "3885219b6810b007914f3a7950a8d1b469d598a5", "url": "https://huggingface.co/unsloth/Qwen3.5-9B-GGUF/tree/3885219b6810b007914f3a7950a8d1b469d598a5", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 921705024, "url": "https://huggingface.co/unsloth/Qwen3.5-9B-GGUF/blob/3885219b6810b007914f3a7950a8d1b469d598a5/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 918166080, "url": "https://huggingface.co/unsloth/Qwen3.5-9B-GGUF/blob/3885219b6810b007914f3a7950a8d1b469d598a5/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1824062016, "url": "https://huggingface.co/unsloth/Qwen3.5-9B-GGUF/blob/3885219b6810b007914f3a7950a8d1b469d598a5/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 6577841376, "repo": "unsloth/Qwen3.5-9B-GGUF", "revision": "3885219b6810b007914f3a7950a8d1b469d598a5", "url": "https://huggingface.co/unsloth/Qwen3.5-9B-GGUF/tree/3885219b6810b007914f3a7950a8d1b469d598a5", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 921705024, "url": "https://huggingface.co/unsloth/Qwen3.5-9B-GGUF/blob/3885219b6810b007914f3a7950a8d1b469d598a5/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 918166080, "url": "https://huggingface.co/unsloth/Qwen3.5-9B-GGUF/blob/3885219b6810b007914f3a7950a8d1b469d598a5/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1824062016, "url": "https://huggingface.co/unsloth/Qwen3.5-9B-GGUF/blob/3885219b6810b007914f3a7950a8d1b469d598a5/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 9527502048, "repo": "unsloth/Qwen3.5-9B-GGUF", "revision": "3885219b6810b007914f3a7950a8d1b469d598a5", "url": "https://huggingface.co/unsloth/Qwen3.5-9B-GGUF/tree/3885219b6810b007914f3a7950a8d1b469d598a5", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 921705024, "url": "https://huggingface.co/unsloth/Qwen3.5-9B-GGUF/blob/3885219b6810b007914f3a7950a8d1b469d598a5/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 918166080, "url": "https://huggingface.co/unsloth/Qwen3.5-9B-GGUF/blob/3885219b6810b007914f3a7950a8d1b469d598a5/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1824062016, "url": "https://huggingface.co/unsloth/Qwen3.5-9B-GGUF/blob/3885219b6810b007914f3a7950a8d1b469d598a5/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 5950221072, "repo": "mlx-community/Qwen3.5-9B-4bit", "revision": "8b2b98c00a6b4d291155e4890773ca8f769aee53", "url": "https://huggingface.co/mlx-community/Qwen3.5-9B-4bit/tree/8b2b98c00a6b4d291155e4890773ca8f769aee53", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 2, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 10426592423, "repo": "mlx-community/Qwen3.5-9B-8bit", "revision": "16daa4818c54ce5f5436f929d52542eb65bbed9d", "url": "https://huggingface.co/mlx-community/Qwen3.5-9B-8bit/tree/16daa4818c54ce5f5436f929d52542eb65bbed9d", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 2, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-09-27", "title": "24-Layer GGUF quants (3.71GB) optimized for 6GB consumer GPUs", "url": "https://huggingface.co/Qwen/Qwen3.5-9B/discussions/68" }, { "platform": "Hugging Face", "date": "2026-05-27", "title": "MTP provides no throughput improvement on Apple Silicon Metal (llama.cpp b9330) — net loss at every configuration", "url": "https://huggingface.co/Qwen/Qwen3.5-9B/discussions/56" }, { "platform": "Reddit", "date": "2026-10-16", "title": null, "url": "https://www.reddit.com/r/LocalLLaMA/comments/1x7g8cc/why_do_we_have_models_under_15b_so_rarely_anymore/" } ], "notes": [] }, { "slug": "exaone-4.5-33b", "name": "EXAONE 4.5 33B", "hf_repo": "LGAI-EXAONE/EXAONE-4.5-33B", "hf_revision": "570aa4b15a4f45ba1133072b45f50198f6e3b4fd", "source_url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B", "org": "LGAI-EXAONE", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-04-09", "precision": "day", "url": "https://github.com/LG-AI-EXAONE/EXAONE-4.5", "status": "verified_primary", "note": "Official dated News release entry." }, "params": { "total_b": 33, "active_b": 33, "moe": false, "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B/blob/570aa4b15a4f45ba1133072b45f50198f6e3b4fd/README.md", "active_url": null, "scope": "Publisher claims 33B including 1.2B vision encoder; HF stored count 34.35B differs." }, "arch_label": "Dense · sliding-window attn", "context": { "native": 262144, "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B/blob/570aa4b15a4f45ba1133072b45f50198f6e3b4fd/config.json", "extended": null }, "license": { "value": "exaone", "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B/blob/570aa4b15a4f45ba1133072b45f50198f6e3b4fd/LICENSE" }, "quality": { "rank": 46, "aa": { "value": 13.2058394009382, "variant": "EXAONE 4.5 33B (Reasoning)", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/exaone-4-5-33b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": 65536, "win_per_token": 196608, "win_tokens": 4096, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B/blob/570aa4b15a4f45ba1133072b45f50198f6e3b4fd/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 20047839424, "repo": "LGAI-EXAONE/EXAONE-4.5-33B-GGUF", "revision": "0e969634ef24db05151b435970297a6dee634b7e", "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B-GGUF/tree/0e969634ef24db05151b435970297a6dee634b7e", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-EXAONE-4.5-33B-BF16.gguf", "bytes": 2579038816, "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B-GGUF/blob/0e969634ef24db05151b435970297a6dee634b7e/mmproj-EXAONE-4.5-33B-BF16.gguf" }, { "path": "mmproj-EXAONE-4.5-33B-F16.gguf", "bytes": 2574221920, "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B-GGUF/blob/0e969634ef24db05151b435970297a6dee634b7e/mmproj-EXAONE-4.5-33B-F16.gguf" }, { "path": "mmproj-EXAONE-4.5-33B-F32.gguf", "bytes": 5145952864, "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B-GGUF/blob/0e969634ef24db05151b435970297a6dee634b7e/mmproj-EXAONE-4.5-33B-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 23482253504, "repo": "LGAI-EXAONE/EXAONE-4.5-33B-GGUF", "revision": "0e969634ef24db05151b435970297a6dee634b7e", "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B-GGUF/tree/0e969634ef24db05151b435970297a6dee634b7e", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-EXAONE-4.5-33B-BF16.gguf", "bytes": 2579038816, "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B-GGUF/blob/0e969634ef24db05151b435970297a6dee634b7e/mmproj-EXAONE-4.5-33B-BF16.gguf" }, { "path": "mmproj-EXAONE-4.5-33B-F16.gguf", "bytes": 2574221920, "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B-GGUF/blob/0e969634ef24db05151b435970297a6dee634b7e/mmproj-EXAONE-4.5-33B-F16.gguf" }, { "path": "mmproj-EXAONE-4.5-33B-F32.gguf", "bytes": 5145952864, "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B-GGUF/blob/0e969634ef24db05151b435970297a6dee634b7e/mmproj-EXAONE-4.5-33B-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 35138742464, "repo": "LGAI-EXAONE/EXAONE-4.5-33B-GGUF", "revision": "0e969634ef24db05151b435970297a6dee634b7e", "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B-GGUF/tree/0e969634ef24db05151b435970297a6dee634b7e", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-EXAONE-4.5-33B-BF16.gguf", "bytes": 2579038816, "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B-GGUF/blob/0e969634ef24db05151b435970297a6dee634b7e/mmproj-EXAONE-4.5-33B-BF16.gguf" }, { "path": "mmproj-EXAONE-4.5-33B-F16.gguf", "bytes": 2574221920, "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B-GGUF/blob/0e969634ef24db05151b435970297a6dee634b7e/mmproj-EXAONE-4.5-33B-F16.gguf" }, { "path": "mmproj-EXAONE-4.5-33B-F32.gguf", "bytes": 5145952864, "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B-GGUF/blob/0e969634ef24db05151b435970297a6dee634b7e/mmproj-EXAONE-4.5-33B-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16", "MLX_4bit", "MLX_8bit" ], "community": [ { "platform": "Hugging Face", "date": "2026-05-04", "title": "vLLM (add-exaone4_5 branch) 빌드 실패 관련 문의드립니다.", "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B/discussions/7" }, { "platform": "Hugging Face", "date": "2026-04-29", "title": "EXAONE 4.5 33B FP8 vLLM ROCm 구동 중 ImageProcessor / ProcessorMixin 정합성 이슈", "url": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.5-33B/discussions/6" } ], "notes": [] }, { "slug": "command-a-plus", "name": "Command A+ (05-2026)", "hf_repo": "CohereLabs/command-a-plus-05-2026-bf16", "hf_revision": "5fb6fde5fd12ff89356aae552e11883bc49f069b", "source_url": "https://huggingface.co/CohereLabs/command-a-plus-05-2026-bf16", "org": "CohereLabs", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-05-20", "precision": "day", "url": "https://cohere.com/blog/cohere-releases-command-a-plus", "status": "verified_primary", "note": "Official announcement and docs changelog; repository created May11 is not release." }, "params": { "total_b": 218, "active_b": 25, "moe": true, "url": "https://huggingface.co/CohereLabs/command-a-plus-05-2026-bf16/raw/5fb6fde5fd12ff89356aae552e11883bc49f069b/README.md", "active_url": "https://huggingface.co/CohereLabs/command-a-plus-05-2026-bf16/raw/5fb6fde5fd12ff89356aae552e11883bc49f069b/README.md", "scope": null }, "arch_label": "MoE · sliding-window attn", "context": { "native": 200000, "url": "https://huggingface.co/CohereLabs/command-a-plus-05-2026-bf16/raw/5fb6fde5fd12ff89356aae552e11883bc49f069b/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/CohereLabs/command-a-plus-05-2026-bf16/blob/5fb6fde5fd12ff89356aae552e11883bc49f069b/LICENSE" }, "quality": { "rank": 47, "aa": { "value": 13.1335002685349, "variant": "Command A+", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/command-a-plus", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": 32768, "win_per_token": null, "win_tokens": null, "win_max": 402653184, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/CohereLabs/command-a-plus-05-2026-bf16/raw/5fb6fde5fd12ff89356aae552e11883bc49f069b/config.json", "dtype": "BF16", "note": "BF16/FP16 KV estimate for full-attention layers only, batch=1. Sliding-window terms and fixed recurrent state are separate; excludes alignment, graphs, buffers, vision and MTP/drafters." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 132266310560, "repo": "DevQuasar/CohereLabs.command-a-plus-05-2026-bf16-GGUF", "revision": "b69a2e4d32289f0f3cccb38a45af81053e38f084", "url": "https://huggingface.co/DevQuasar/CohereLabs.command-a-plus-05-2026-bf16-GGUF/tree/b69a2e4d32289f0f3cccb38a45af81053e38f084", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 9, "status": "measured_file_manifest" }, "GGUF_Q5_K_M": { "bytes": 154974272672, "repo": "DevQuasar/CohereLabs.command-a-plus-05-2026-bf16-GGUF", "revision": "b69a2e4d32289f0f3cccb38a45af81053e38f084", "url": "https://huggingface.co/DevQuasar/CohereLabs.command-a-plus-05-2026-bf16-GGUF/tree/b69a2e4d32289f0f3cccb38a45af81053e38f084", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 11, "status": "measured_file_manifest" }, "GGUF_Q8_0": { "bytes": 231956004448, "repo": "DevQuasar/CohereLabs.command-a-plus-05-2026-bf16-GGUF", "revision": "b69a2e4d32289f0f3cccb38a45af81053e38f084", "url": "https://huggingface.co/DevQuasar/CohereLabs.command-a-plus-05-2026-bf16-GGUF/tree/b69a2e4d32289f0f3cccb38a45af81053e38f084", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 16, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_FP16", "GGUF_BF16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "qwen3.5-4b", "name": "Qwen3.5-4B", "hf_repo": "Qwen/Qwen3.5-4B", "hf_revision": "851bf6e806efd8d0a36b00ddf55e13ccb7b8cd0a", "source_url": "https://huggingface.co/Qwen/Qwen3.5-4B", "org": "Qwen", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-03-02", "precision": "day", "url": "https://github.com/QwenLM/Qwen3.8/blob/main/README.md", "status": "verified_primary", "note": "Official dated release-news entry states public model availability." }, "params": { "total_b": 4, "active_b": 4, "moe": false, "url": "https://huggingface.co/Qwen/Qwen3.5-4B/blob/851bf6e806efd8d0a36b00ddf55e13ccb7b8cd0a/README.md", "active_url": null, "scope": "Publisher nominal language-model size; HF total includes stored multimodal/MTP tensors as applicable." }, "arch_label": "Dense · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/Qwen/Qwen3.5-4B/blob/851bf6e806efd8d0a36b00ddf55e13ccb7b8cd0a/config.json", "extended": 1010000 }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/Qwen/Qwen3.5-4B/blob/main/LICENSE" }, "quality": { "rank": 48, "aa": { "value": 13.1248174602542, "variant": "Qwen3.5 4B (Reasoning)", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/qwen3-5-4b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": 32768, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3.5-4B/blob/851bf6e806efd8d0a36b00ddf55e13ccb7b8cd0a/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 2740937888, "repo": "unsloth/Qwen3.5-4B-GGUF", "revision": "e87f176479d0855a907a41277aca2f8ee7a09523", "url": "https://huggingface.co/unsloth/Qwen3.5-4B-GGUF/tree/e87f176479d0855a907a41277aca2f8ee7a09523", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 675569344, "url": "https://huggingface.co/unsloth/Qwen3.5-4B-GGUF/blob/e87f176479d0855a907a41277aca2f8ee7a09523/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 672423616, "url": "https://huggingface.co/unsloth/Qwen3.5-4B-GGUF/blob/e87f176479d0855a907a41277aca2f8ee7a09523/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1334075072, "url": "https://huggingface.co/unsloth/Qwen3.5-4B-GGUF/blob/e87f176479d0855a907a41277aca2f8ee7a09523/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 3143656608, "repo": "unsloth/Qwen3.5-4B-GGUF", "revision": "e87f176479d0855a907a41277aca2f8ee7a09523", "url": "https://huggingface.co/unsloth/Qwen3.5-4B-GGUF/tree/e87f176479d0855a907a41277aca2f8ee7a09523", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 675569344, "url": "https://huggingface.co/unsloth/Qwen3.5-4B-GGUF/blob/e87f176479d0855a907a41277aca2f8ee7a09523/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 672423616, "url": "https://huggingface.co/unsloth/Qwen3.5-4B-GGUF/blob/e87f176479d0855a907a41277aca2f8ee7a09523/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1334075072, "url": "https://huggingface.co/unsloth/Qwen3.5-4B-GGUF/blob/e87f176479d0855a907a41277aca2f8ee7a09523/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 4482403488, "repo": "unsloth/Qwen3.5-4B-GGUF", "revision": "e87f176479d0855a907a41277aca2f8ee7a09523", "url": "https://huggingface.co/unsloth/Qwen3.5-4B-GGUF/tree/e87f176479d0855a907a41277aca2f8ee7a09523", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 675569344, "url": "https://huggingface.co/unsloth/Qwen3.5-4B-GGUF/blob/e87f176479d0855a907a41277aca2f8ee7a09523/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 672423616, "url": "https://huggingface.co/unsloth/Qwen3.5-4B-GGUF/blob/e87f176479d0855a907a41277aca2f8ee7a09523/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1334075072, "url": "https://huggingface.co/unsloth/Qwen3.5-4B-GGUF/blob/e87f176479d0855a907a41277aca2f8ee7a09523/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 3034300695, "repo": "mlx-community/Qwen3.5-4B-4bit", "revision": "0e7ffd5c629ef7719d4cbc04069232580bfa9d9c", "url": "https://huggingface.co/mlx-community/Qwen3.5-4B-4bit/tree/0e7ffd5c629ef7719d4cbc04069232580bfa9d9c", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 5136696107, "repo": "mlx-community/Qwen3.5-4B-8bit", "revision": "1704583a86670ba531b7f2c92530a2f725cfd45b", "url": "https://huggingface.co/mlx-community/Qwen3.5-4B-8bit/tree/1704583a86670ba531b7f2c92530a2f725cfd45b", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-03-06", "title": "openai.APIConnectionError: Connection error.", "url": "https://huggingface.co/Qwen/Qwen3.5-4B/discussions/5" }, { "platform": "Hugging Face", "date": "2026-10-14", "title": "Request: DOI", "url": "https://huggingface.co/Qwen/Qwen3.5-4B/discussions/30" } ], "notes": [] }, { "slug": "nemotron-3.5-lightning-30b-a3b", "name": "NVIDIA Nemotron 3.5 Lightning 30B-A3B", "hf_repo": "nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16", "hf_revision": "a9904d24bcc1d289a1950fa9d2b978c47cf903b9", "source_url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16", "org": "nvidia", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08-11", "precision": "day", "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16/blob/a9904d24bcc1d289a1950fa9d2b978c47cf903b9/README.md", "status": "verified_primary", "note": "Model card explicit Release Date." }, "params": { "total_b": 30, "active_b": 3, "moe": true, "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16/blob/a9904d24bcc1d289a1950fa9d2b978c47cf903b9/README.md", "active_url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16/blob/a9904d24bcc1d289a1950fa9d2b978c47cf903b9/README.md", "scope": "Publisher nominal total/active counts." }, "arch_label": "MoE · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16/blob/a9904d24bcc1d289a1950fa9d2b978c47cf903b9/config.json", "extended": 1048576 }, "license": { "value": "openmdw-1.1", "url": "https://openmdw.ai/license/1-1/" }, "quality": { "rank": 49, "aa": { "value": 12.8572306217144, "variant": "Nemotron 3.5 Lightning", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/nemotron-3-5-lightning", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1328.8829530844, "lower": 1323.3529122803, "upper": 1334.4129938885, "votes": 19938, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": 6144, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16/blob/a9904d24bcc1d289a1950fa9d2b978c47cf903b9/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 25477403616, "repo": "bartowski/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF", "revision": "f0eec2267ae843d9eb21ea3926ab0046da0a8628", "url": "https://huggingface.co/bartowski/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF/tree/f0eec2267ae843d9eb21ea3926ab0046da0a8628", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 26964179424, "repo": "bartowski/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF", "revision": "f0eec2267ae843d9eb21ea3926ab0046da0a8628", "url": "https://huggingface.co/bartowski/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF/tree/f0eec2267ae843d9eb21ea3926ab0046da0a8628", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 35004643296, "repo": "bartowski/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF", "revision": "f0eec2267ae843d9eb21ea3926ab0046da0a8628", "url": "https://huggingface.co/bartowski/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF/tree/f0eec2267ae843d9eb21ea3926ab0046da0a8628", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 17775425359, "repo": "mlx-community/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-4bit", "revision": "55ac8c89261109b36c04371cd3f479a4594208c8", "url": "https://huggingface.co/mlx-community/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-4bit/tree/55ac8c89261109b36c04371cd3f479a4594208c8", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 4, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 33559963550, "repo": "mlx-community/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-8bit", "revision": "881a85f9e84dac5ab1a61d4493739b5ef67efeaa", "url": "https://huggingface.co/mlx-community/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-8bit/tree/881a85f9e84dac5ab1a61d4493739b5ef67efeaa", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 7, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-08-24", "title": "Benchmark: fine-tuned Nemotron 3 Nano vs. nine frontier LLMs on B2B propensity ranking", "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16/discussions/10" }, { "platform": "Hugging Face", "date": "2026-08-19", "title": "RTX PRO 6000 96GB", "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16/discussions/8" } ], "notes": [] }, { "slug": "granite-4.2-30b", "name": "IBM Granite 4.2 30B", "hf_repo": "ibm-granite/granite-4.2-30b", "hf_revision": "9e668ce1c538387ef24d3644e9b0606647762636", "source_url": "https://huggingface.co/ibm-granite/granite-4.2-30b", "org": "ibm-granite", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08-25", "precision": "day", "url": "https://huggingface.co/ibm-granite/granite-4.2-30b/blob/9e668ce1c538387ef24d3644e9b0606647762636/README.md", "status": "verified_primary", "note": "Model card explicit Release Date." }, "params": { "total_b": 30, "active_b": 30, "moe": false, "url": "https://huggingface.co/ibm-granite/granite-4.2-30b/blob/9e668ce1c538387ef24d3644e9b0606647762636/README.md", "active_url": null, "scope": "Publisher nominal size." }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/ibm-granite/granite-4.2-30b/blob/9e668ce1c538387ef24d3644e9b0606647762636/config.json", "extended": 524288 }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/ibm-granite/granite-4.2-30b/blob/9e668ce1c538387ef24d3644e9b0606647762636/LICENSE" }, "quality": { "rank": 50, "aa": { "value": 12.68315341158, "variant": "Granite 4.2 30B", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/granite-4-2-30b", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1361.4062159323, "lower": 1350.9777286916, "upper": 1371.834703173, "votes": 3401, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": 262144, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/ibm-granite/granite-4.2-30b/blob/9e668ce1c538387ef24d3644e9b0606647762636/config.json", "dtype": "FP16", "note": "Ordinary GQA cache estimate; actual allocation and padding vary by backend." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 17721455328, "repo": "ibm-granite/granite-4.2-30b-GGUF", "revision": "27b350a791e81d9a4d1ddca1c49282e9ec533768", "url": "https://huggingface.co/ibm-granite/granite-4.2-30b-GGUF/tree/27b350a791e81d9a4d1ddca1c49282e9ec533768", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 20775957216, "repo": "ibm-granite/granite-4.2-30b-GGUF", "revision": "27b350a791e81d9a4d1ddca1c49282e9ec533768", "url": "https://huggingface.co/ibm-granite/granite-4.2-30b-GGUF/tree/27b350a791e81d9a4d1ddca1c49282e9ec533768", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 31111705312, "repo": "ibm-granite/granite-4.2-30b-GGUF", "revision": "27b350a791e81d9a4d1ddca1c49282e9ec533768", "url": "https://huggingface.co/ibm-granite/granite-4.2-30b-GGUF/tree/27b350a791e81d9a4d1ddca1c49282e9ec533768", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 16469111945, "repo": "lmstudio-community/granite-4.2-30b-MLX-4bit", "revision": "a9b33b84944a3c27062c60bad1a37e0299f04eb2", "url": "https://huggingface.co/lmstudio-community/granite-4.2-30b-MLX-4bit/tree/a9b33b84944a3c27062c60bad1a37e0299f04eb2", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 4, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 31107233472, "repo": "lmstudio-community/granite-4.2-30b-MLX-8bit", "revision": "f64edd9dadd896fe1ca924c3c35be10509f674df", "url": "https://huggingface.co/lmstudio-community/granite-4.2-30b-MLX-8bit/tree/f64edd9dadd896fe1ca924c3c35be10509f674df", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 6, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-09-03", "title": "Thinking dilemmas", "url": "https://huggingface.co/ibm-granite/granite-4.2-30b/discussions/5" }, { "platform": "Hugging Face", "date": "2026-09-01", "title": "Thank you for the contribution", "url": "https://huggingface.co/ibm-granite/granite-4.2-30b/discussions/4" } ], "notes": [] }, { "slug": "minicpm5-2b", "name": "MiniCPM5-2B", "hf_repo": "openbmb/MiniCPM5-2B", "hf_revision": "f97400052a43d642bbc6e9975e2397e3ae6a6b52", "source_url": "https://huggingface.co/openbmb/MiniCPM5-2B", "org": "openbmb", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-09-07", "precision": "day", "url": "https://github.com/OpenBMB/MiniCPM", "status": "verified_primary", "note": "Official release changelog; not September 6 repository creation." }, "params": { "total_b": 2.51675648, "active_b": 2.51675648, "moe": false, "url": "https://huggingface.co/openbmb/MiniCPM5-2B/blob/f97400052a43d642bbc6e9975e2397e3ae6a6b52/README.md", "active_url": null, "scope": "Publisher exact total; non-embedding count 1,981,982,720." }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/openbmb/MiniCPM5-2B/blob/f97400052a43d642bbc6e9975e2397e3ae6a6b52/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/openbmb/MiniCPM5-2B/blob/f97400052a43d642bbc6e9975e2397e3ae6a6b52/LICENSE" }, "quality": { "rank": 51, "aa": { "value": 12.463354831996, "variant": "MiniCPM5-2B", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/minicpm5-2b", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": 43008, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/openbmb/MiniCPM5-2B/blob/f97400052a43d642bbc6e9975e2397e3ae6a6b52/config.json", "dtype": "FP16", "note": "Ordinary GQA cache estimate; actual allocation and padding vary by backend." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 1561318368, "repo": "openbmb/MiniCPM5-2B-GGUF", "revision": "2079a22f3beaa4e306449978533478fe0522f4b3", "url": "https://huggingface.co/openbmb/MiniCPM5-2B-GGUF/tree/2079a22f3beaa4e306449978533478fe0522f4b3", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 1920175328, "repo": "bartowski/MiniCPM5-2B-GGUF", "revision": "31fee0334c4ed9f1e079afe586ba30cb83e0f883", "url": "https://huggingface.co/bartowski/MiniCPM5-2B-GGUF/tree/31fee0334c4ed9f1e079afe586ba30cb83e0f883", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 2679710688, "repo": "openbmb/MiniCPM5-2B-GGUF", "revision": "2079a22f3beaa4e306449978533478fe0522f4b3", "url": "https://huggingface.co/openbmb/MiniCPM5-2B-GGUF/tree/2079a22f3beaa4e306449978533478fe0522f4b3", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 5039006688, "repo": "openbmb/MiniCPM5-2B-GGUF", "revision": "2079a22f3beaa4e306449978533478fe0522f4b3", "url": "https://huggingface.co/openbmb/MiniCPM5-2B-GGUF/tree/2079a22f3beaa4e306449978533478fe0522f4b3", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 1416035216, "repo": "openbmb/MiniCPM5-2B-MLX", "revision": "8a9ad7539ac86281d0ac2b017ba04a5de53fe9a3", "url": "https://huggingface.co/openbmb/MiniCPM5-2B-MLX/tree/8a9ad7539ac86281d0ac2b017ba04a5de53fe9a3", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 2674327290, "repo": "mlx-community/MiniCPM5-2B-8bit", "revision": "2d20e8e672ce892d50f7265bfd3fc9b59b718f2a", "url": "https://huggingface.co/mlx-community/MiniCPM5-2B-8bit/tree/2d20e8e672ce892d50f7265bfd3fc9b59b718f2a", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-09-26", "title": "MiniCPM5-2B-Q8_0_XL-GGUF", "url": "https://huggingface.co/openbmb/MiniCPM5-2B/discussions/20" }, { "platform": "Hugging Face", "date": "2026-09-08", "title": "it will be good if model supports turkish", "url": "https://huggingface.co/openbmb/MiniCPM5-2B/discussions/8" } ], "notes": [] }, { "slug": "hypernova-60b-2605", "name": "HyperNova 60B 2605 (compressed gpt-oss-120b)", "hf_repo": "MultiverseComputingCAI/Hypernova-60B-2605", "hf_revision": "373c5c748c14e779e8edc21ea4e8cb7fb26e5e8a", "source_url": "https://huggingface.co/MultiverseComputingCAI/Hypernova-60B-2605", "org": "MultiverseComputingCAI", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-05-06", "precision": "day", "url": "https://multiversecomputing.com/resources/introducing-hypernova-60b-2605", "status": "verified_primary", "note": "Model-specific 2605 announcement. HF card table retains Feb26 date from earlier 2602 version." }, "params": { "total_b": 60, "active_b": 4.8, "moe": true, "url": "https://huggingface.co/MultiverseComputingCAI/Hypernova-60B-2605/raw/373c5c748c14e779e8edc21ea4e8cb7fb26e5e8a/README.md", "active_url": "https://huggingface.co/MultiverseComputingCAI/Hypernova-60B-2605/raw/373c5c748c14e779e8edc21ea4e8cb7fb26e5e8a/README.md", "scope": null }, "arch_label": "MoE · sliding-window attn", "context": { "native": 131072, "url": "https://huggingface.co/MultiverseComputingCAI/Hypernova-60B-2605/raw/373c5c748c14e779e8edc21ea4e8cb7fb26e5e8a/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/MultiverseComputingCAI/Hypernova-60B-2605/blob/373c5c748c14e779e8edc21ea4e8cb7fb26e5e8a/LICENSE" }, "quality": { "rank": 52, "aa": { "value": 11.7380211033452, "variant": "HyperNova 60B 2605 (High, Based on gpt-oss-120b)", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/hypernova-60b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": 32768, "win_per_token": null, "win_tokens": null, "win_max": 4194304, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/MultiverseComputingCAI/Hypernova-60B-2605/raw/373c5c748c14e779e8edc21ea4e8cb7fb26e5e8a/config.json", "dtype": "BF16", "note": "BF16/FP16 KV estimate for full-attention layers only, batch=1. Sliding-window terms and fixed recurrent state are separate; excludes alignment, graphs, buffers, vision and MTP/drafters." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 40687403968, "repo": "mradermacher/Hypernova-60B-2605-i1-GGUF", "revision": "03d8cfcdbb0202224ab71e2b30ea3d3e6638592e", "url": "https://huggingface.co/mradermacher/Hypernova-60B-2605-i1-GGUF/tree/03d8cfcdbb0202224ab71e2b30ea3d3e6638592e", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 1, "status": "measured_file_manifest" }, "GGUF_Q5_K_M": { "bytes": 44337746368, "repo": "mradermacher/Hypernova-60B-2605-i1-GGUF", "revision": "03d8cfcdbb0202224ab71e2b30ea3d3e6638592e", "url": "https://huggingface.co/mradermacher/Hypernova-60B-2605-i1-GGUF/tree/03d8cfcdbb0202224ab71e2b30ea3d3e6638592e", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 1, "status": "measured_file_manifest" }, "GGUF_Q8_0": { "bytes": 62421358272, "repo": "mradermacher/Hypernova-60B-2605-GGUF", "revision": "5c7638d5b498210d7ca8bc03fb5499c29435a670", "url": "https://huggingface.co/mradermacher/Hypernova-60B-2605-GGUF/tree/5c7638d5b498210d7ca8bc03fb5499c29435a670", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 1, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_FP16", "GGUF_BF16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [ "Card release table says 26/02/2026 while model-specific official 2605 release announcement says May6; keep conflict. Native MXFP4 compression." ] }, { "slug": "nemotron-cascade-2-30b-a3b", "name": "Nemotron Cascade 2 30B A3B", "hf_repo": "nvidia/Nemotron-Cascade-2-30B-A3B", "hf_revision": "6327cdbcf907e1c7cec9cb29fb6e6cebdf8feaf7", "source_url": "https://huggingface.co/nvidia/Nemotron-Cascade-2-30B-A3B", "org": "nvidia", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-03-19", "precision": "day", "url": "https://huggingface.co/nvidia/Nemotron-Cascade-2-30B-A3B/blob/6327cdbcf907e1c7cec9cb29fb6e6cebdf8feaf7/README.md", "status": "verified_primary", "note": "Model card explicit Release Date." }, "params": { "total_b": 30, "active_b": 3, "moe": true, "url": "https://huggingface.co/nvidia/Nemotron-Cascade-2-30B-A3B/blob/6327cdbcf907e1c7cec9cb29fb6e6cebdf8feaf7/README.md", "active_url": "https://huggingface.co/nvidia/Nemotron-Cascade-2-30B-A3B/blob/6327cdbcf907e1c7cec9cb29fb6e6cebdf8feaf7/README.md", "scope": "Publisher nominal total/active counts." }, "arch_label": "MoE · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/nvidia/Nemotron-Cascade-2-30B-A3B/blob/6327cdbcf907e1c7cec9cb29fb6e6cebdf8feaf7/config.json", "extended": null }, "license": { "value": "nvidia-open-model-license", "url": "https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license/" }, "quality": { "rank": 53, "aa": { "value": 11.6708359406601, "variant": "Nemotron Cascade 2 30B A3B", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/nemotron-cascade-2-30b-a3b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": 6144, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/nvidia/Nemotron-Cascade-2-30B-A3B/blob/6327cdbcf907e1c7cec9cb29fb6e6cebdf8feaf7/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 24515130976, "repo": "mradermacher/Nemotron-Cascade-2-30B-A3B-GGUF", "revision": "d27b10b50877cdb55c38deb5e0f4d7eb6c55f6cc", "url": "https://huggingface.co/mradermacher/Nemotron-Cascade-2-30B-A3B-GGUF/tree/d27b10b50877cdb55c38deb5e0f4d7eb6c55f6cc", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 26040162400, "repo": "mradermacher/Nemotron-Cascade-2-30B-A3B-GGUF", "revision": "d27b10b50877cdb55c38deb5e0f4d7eb6c55f6cc", "url": "https://huggingface.co/mradermacher/Nemotron-Cascade-2-30B-A3B-GGUF/tree/d27b10b50877cdb55c38deb5e0f4d7eb6c55f6cc", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 33585496672, "repo": "mradermacher/Nemotron-Cascade-2-30B-A3B-GGUF", "revision": "d27b10b50877cdb55c38deb5e0f4d7eb6c55f6cc", "url": "https://huggingface.co/mradermacher/Nemotron-Cascade-2-30B-A3B-GGUF/tree/d27b10b50877cdb55c38deb5e0f4d7eb6c55f6cc", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 63181505152, "repo": "freddm/Nemotron-Cascade-2-30B-A3B-GGUF", "revision": "3214f30fbe38322e10c64d2ba5c06bb381ae935d", "url": "https://huggingface.co/freddm/Nemotron-Cascade-2-30B-A3B-GGUF/tree/3214f30fbe38322e10c64d2ba5c06bb381ae935d", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 17775425379, "repo": "mlx-community/Nemotron-Cascade-2-30B-A3B-4bit", "revision": "cf40185ae9f11417eef7f9af6f6d5e11740e52a6", "url": "https://huggingface.co/mlx-community/Nemotron-Cascade-2-30B-A3B-4bit/tree/cf40185ae9f11417eef7f9af6f6d5e11740e52a6", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 4, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 33559963166, "repo": "mlx-community/Nemotron-Cascade-2-30B-A3B-8bit", "revision": "c39b12797af76d51b1203683ab48274244fb14eb", "url": "https://huggingface.co/mlx-community/Nemotron-Cascade-2-30B-A3B-8bit/tree/c39b12797af76d51b1203683ab48274244fb14eb", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 7, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-03-31", "title": "Nemotron Cascade mutates nested path literals in tool arguments even with explicit literal-preservation instructions and temperature=0.", "url": "https://huggingface.co/nvidia/Nemotron-Cascade-2-30B-A3B/discussions/23" }, { "platform": "Hugging Face", "date": "2026-03-25", "title": "modeling_nemotron_h.py: Multiple bugs in HybridMambaAttentionDynamicCache break generation with CUDA fast path", "url": "https://huggingface.co/nvidia/Nemotron-Cascade-2-30B-A3B/discussions/13" } ], "notes": [] }, { "slug": "longcat-flash-lite", "name": "LongCat-Flash-Lite", "hf_repo": "meituan-longcat/LongCat-Flash-Lite", "hf_revision": "b62b68827ead0b7fef3ba98b57f18484acaaec06", "source_url": "https://huggingface.co/meituan-longcat/LongCat-Flash-Lite", "org": "meituan-longcat", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-01", "precision": "month", "url": "https://arxiv.org/abs/2601.21204", "status": "verified_publisher_release_month", "note": "Publisher January2026 paper introduces released model; exact weight day unverified." }, "params": { "total_b": 68.5, "active_b": 3, "moe": true, "url": "https://huggingface.co/meituan-longcat/LongCat-Flash-Lite/raw/b62b68827ead0b7fef3ba98b57f18484acaaec06/README.md", "active_url": "https://huggingface.co/meituan-longcat/LongCat-Flash-Lite/raw/b62b68827ead0b7fef3ba98b57f18484acaaec06/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 327680, "url": "https://huggingface.co/meituan-longcat/LongCat-Flash-Lite/raw/b62b68827ead0b7fef3ba98b57f18484acaaec06/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/meituan-longcat/LongCat-Flash-Lite/blob/b62b68827ead0b7fef3ba98b57f18484acaaec06/LICENSE" }, "quality": { "rank": 54, "aa": { "value": 11.4709079965186, "variant": "LongCat Flash Lite", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/longcat-flash-lite", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/meituan-longcat/LongCat-Flash-Lite/raw/b62b68827ead0b7fef3ba98b57f18484acaaec06/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 40231409728, "repo": "InquiringMinds-AI/LongCat-Flash-Lite-GGUF", "revision": "43130dbea82a0daf7bac400e0084f1d780d2cd67", "url": "https://huggingface.co/InquiringMinds-AI/LongCat-Flash-Lite-GGUF/tree/43130dbea82a0daf7bac400e0084f1d780d2cd67", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 1, "status": "measured_file_manifest" }, "GGUF_Q5_K_M": { "bytes": 48007948864, "repo": "InquiringMinds-AI/LongCat-Flash-Lite-GGUF", "revision": "43130dbea82a0daf7bac400e0084f1d780d2cd67", "url": "https://huggingface.co/InquiringMinds-AI/LongCat-Flash-Lite-GGUF/tree/43130dbea82a0daf7bac400e0084f1d780d2cd67", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 1, "status": "measured_file_manifest" }, "GGUF_Q8_0": { "bytes": 72871665728, "repo": "InquiringMinds-AI/LongCat-Flash-Lite-GGUF", "revision": "43130dbea82a0daf7bac400e0084f1d780d2cd67", "url": "https://huggingface.co/InquiringMinds-AI/LongCat-Flash-Lite-GGUF/tree/43130dbea82a0daf7bac400e0084f1d780d2cd67", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 1, "status": "measured_file_manifest" }, "GGUF_BF16": { "bytes": 137133270656, "repo": "InquiringMinds-AI/LongCat-Flash-Lite-GGUF", "revision": "43130dbea82a0daf7bac400e0084f1d780d2cd67", "url": "https://huggingface.co/InquiringMinds-AI/LongCat-Flash-Lite-GGUF/tree/43130dbea82a0daf7bac400e0084f1d780d2cd67", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 1, "status": "measured_file_manifest" }, "MLX_4bit": { "bytes": 38575121528, "repo": "mlx-community/LongCat-Flash-Lite-4bit", "revision": "0435b3bca4418830335d784317fe7f67f8383253", "url": "https://huggingface.co/mlx-community/LongCat-Flash-Lite-4bit/tree/0435b3bca4418830335d784317fe7f67f8383253", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 8, "status": "measured_file_manifest", "mixed": true }, "MLX_8bit": { "bytes": 72847977596, "repo": "mlx-community/LongCat-Flash-Lite-8bit", "revision": "ad91ece142e56d126cbadbd1c93eb5dd31bafcae", "url": "https://huggingface.co/mlx-community/LongCat-Flash-Lite-8bit/tree/ad91ece142e56d126cbadbd1c93eb5dd31bafcae", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 20, "status": "measured_file_manifest", "mixed": true } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [], "notes": [ "Card also reports 2.9B–4.5B active depending on token routing; over 30B n-gram embeddings." ] }, { "slug": "mistral-small-4", "name": "Mistral Small 4 (119B-2603)", "hf_repo": "mistralai/Mistral-Small-4-119B-2603", "hf_revision": "a11f36bebf709121056b1dbcc943d1c6afbe494d", "source_url": "https://huggingface.co/mistralai/Mistral-Small-4-119B-2603", "org": "mistralai", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-03-16", "precision": "day", "url": "https://mistral.ai/news/mistral-small-4/", "status": "verified_primary", "note": null }, "params": { "total_b": 119, "active_b": 6, "moe": true, "url": "https://huggingface.co/mistralai/Mistral-Small-4-119B-2603/raw/a11f36bebf709121056b1dbcc943d1c6afbe494d/README.md", "active_url": "https://huggingface.co/mistralai/Mistral-Small-4-119B-2603/raw/a11f36bebf709121056b1dbcc943d1c6afbe494d/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/mistralai/Mistral-Small-4-119B-2603/raw/a11f36bebf709121056b1dbcc943d1c6afbe494d/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/mistralai/Mistral-Small-4-119B-2603/blob/a11f36bebf709121056b1dbcc943d1c6afbe494d/LICENSE" }, "quality": { "rank": 55, "aa": { "value": 11.2675794141469, "variant": "Mistral Small 4 (Reasoning)", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/mistral-small-4", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/mistralai/Mistral-Small-4-119B-2603/raw/a11f36bebf709121056b1dbcc943d1c6afbe494d/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 72636771776, "repo": "bartowski/mistralai_Mistral-Small-4-119B-2603-GGUF", "revision": "f569d04efea9546de10ec040819908fa4e8f1c92", "url": "https://huggingface.co/bartowski/mistralai_Mistral-Small-4-119B-2603-GGUF/tree/f569d04efea9546de10ec040819908fa4e8f1c92", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 2, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-mistralai_Mistral-Small-4-119B-2603-bf16.gguf", "bytes": 866670848, "url": "https://huggingface.co/bartowski/mistralai_Mistral-Small-4-119B-2603-GGUF/resolve/f569d04efea9546de10ec040819908fa4e8f1c92/mmproj-mistralai_Mistral-Small-4-119B-2603-bf16.gguf" }, { "path": "mmproj-mistralai_Mistral-Small-4-119B-2603-f16.gguf", "bytes": 857078016, "url": "https://huggingface.co/bartowski/mistralai_Mistral-Small-4-119B-2603-GGUF/resolve/f569d04efea9546de10ec040819908fa4e8f1c92/mmproj-mistralai_Mistral-Small-4-119B-2603-f16.gguf" } ] }, "GGUF_Q5_K_M": { "bytes": 85039164992, "repo": "bartowski/mistralai_Mistral-Small-4-119B-2603-GGUF", "revision": "f569d04efea9546de10ec040819908fa4e8f1c92", "url": "https://huggingface.co/bartowski/mistralai_Mistral-Small-4-119B-2603-GGUF/tree/f569d04efea9546de10ec040819908fa4e8f1c92", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 3, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-mistralai_Mistral-Small-4-119B-2603-bf16.gguf", "bytes": 866670848, "url": "https://huggingface.co/bartowski/mistralai_Mistral-Small-4-119B-2603-GGUF/resolve/f569d04efea9546de10ec040819908fa4e8f1c92/mmproj-mistralai_Mistral-Small-4-119B-2603-bf16.gguf" }, { "path": "mmproj-mistralai_Mistral-Small-4-119B-2603-f16.gguf", "bytes": 857078016, "url": "https://huggingface.co/bartowski/mistralai_Mistral-Small-4-119B-2603-GGUF/resolve/f569d04efea9546de10ec040819908fa4e8f1c92/mmproj-mistralai_Mistral-Small-4-119B-2603-f16.gguf" } ] }, "GGUF_Q8_0": { "bytes": 126472990400, "repo": "bartowski/mistralai_Mistral-Small-4-119B-2603-GGUF", "revision": "f569d04efea9546de10ec040819908fa4e8f1c92", "url": "https://huggingface.co/bartowski/mistralai_Mistral-Small-4-119B-2603-GGUF/tree/f569d04efea9546de10ec040819908fa4e8f1c92", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 4, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-mistralai_Mistral-Small-4-119B-2603-bf16.gguf", "bytes": 866670848, "url": "https://huggingface.co/bartowski/mistralai_Mistral-Small-4-119B-2603-GGUF/resolve/f569d04efea9546de10ec040819908fa4e8f1c92/mmproj-mistralai_Mistral-Small-4-119B-2603-bf16.gguf" }, { "path": "mmproj-mistralai_Mistral-Small-4-119B-2603-f16.gguf", "bytes": 857078016, "url": "https://huggingface.co/bartowski/mistralai_Mistral-Small-4-119B-2603-GGUF/resolve/f569d04efea9546de10ec040819908fa4e8f1c92/mmproj-mistralai_Mistral-Small-4-119B-2603-f16.gguf" } ] }, "GGUF_BF16": { "bytes": 237992005728, "repo": "unsloth/Mistral-Small-4-119B-2603-GGUF", "revision": "bd93c721735aa32c035c0f19e738cb3371fd56ff", "url": "https://huggingface.co/unsloth/Mistral-Small-4-119B-2603-GGUF/tree/bd93c721735aa32c035c0f19e738cb3371fd56ff", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 6, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 866671328, "url": "https://huggingface.co/unsloth/Mistral-Small-4-119B-2603-GGUF/resolve/bd93c721735aa32c035c0f19e738cb3371fd56ff/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 857078496, "url": "https://huggingface.co/unsloth/Mistral-Small-4-119B-2603-GGUF/resolve/bd93c721735aa32c035c0f19e738cb3371fd56ff/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1713920736, "url": "https://huggingface.co/unsloth/Mistral-Small-4-119B-2603-GGUF/resolve/bd93c721735aa32c035c0f19e738cb3371fd56ff/mmproj-F32.gguf" } ] } }, "artifacts_missing": [ "GGUF_FP16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [ "Card/current config and original launch context differ; preserve both and require runtime validation. Active size convention varies: 6B excludes embeddings/output, 8B includes them.", "Official launch convention excludes embedding/output parameters from active count; other sources include them and say 8B." ] }, { "slug": "granite-4.2-8b", "name": "Granite 4.2 8B", "hf_repo": "ibm-granite/granite-4.2-8b", "hf_revision": "f8de16cdcdbc6c779ca517604e050d82cc119e44", "source_url": "https://huggingface.co/ibm-granite/granite-4.2-8b", "org": "ibm-granite", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08-25", "precision": "day", "url": "https://huggingface.co/ibm-granite/granite-4.2-8b/blob/f8de16cdcdbc6c779ca517604e050d82cc119e44/README.md", "status": "verified_primary", "note": "Model card explicit Release Date." }, "params": { "total_b": 8, "active_b": 8, "moe": false, "url": "https://huggingface.co/ibm-granite/granite-4.2-8b/blob/f8de16cdcdbc6c779ca517604e050d82cc119e44/README.md", "active_url": null, "scope": "Publisher nominal size." }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/ibm-granite/granite-4.2-8b/blob/f8de16cdcdbc6c779ca517604e050d82cc119e44/config.json", "extended": 524288 }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/ibm-granite/granite-4.2-8b/blob/f8de16cdcdbc6c779ca517604e050d82cc119e44/LICENSE" }, "quality": { "rank": 56, "aa": { "value": 11.0812712267815, "variant": "Granite 4.2 8B", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/granite-4-2-8b", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1319.5785614364, "lower": 1308.5112778011, "upper": 1330.6458450718, "votes": 3254, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": 163840, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/ibm-granite/granite-4.2-8b/blob/f8de16cdcdbc6c779ca517604e050d82cc119e44/config.json", "dtype": "FP16", "note": "Ordinary GQA cache estimate; actual allocation and padding vary by backend." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 5347917952, "repo": "ibm-granite/granite-4.2-8b-GGUF", "revision": "93f3f6a8938ee922b784cf4e5b4203cd3428df8f", "url": "https://huggingface.co/ibm-granite/granite-4.2-8b-GGUF/tree/93f3f6a8938ee922b784cf4e5b4203cd3428df8f", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 6253887616, "repo": "ibm-granite/granite-4.2-8b-GGUF", "revision": "93f3f6a8938ee922b784cf4e5b4203cd3428df8f", "url": "https://huggingface.co/ibm-granite/granite-4.2-8b-GGUF/tree/93f3f6a8938ee922b784cf4e5b4203cd3428df8f", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 9345613952, "repo": "ibm-granite/granite-4.2-8b-GGUF", "revision": "93f3f6a8938ee922b784cf4e5b4203cd3428df8f", "url": "https://huggingface.co/ibm-granite/granite-4.2-8b-GGUF/tree/93f3f6a8938ee922b784cf4e5b4203cd3428df8f", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 17587421760, "repo": "mradermacher/granite-4.2-8b-GGUF", "revision": "72e0876f143f0642399a9a0b2565b697706d7c27", "url": "https://huggingface.co/mradermacher/granite-4.2-8b-GGUF/tree/72e0876f143f0642399a9a0b2565b697706d7c27", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 4945853970, "repo": "lmstudio-community/granite-4.2-8b-MLX-4bit", "revision": "1cfd523c66d94e2697af82cd968368d050166af4", "url": "https://huggingface.co/lmstudio-community/granite-4.2-8b-MLX-4bit/tree/1cfd523c66d94e2697af82cd968368d050166af4", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 9341484710, "repo": "lmstudio-community/granite-4.2-8b-MLX-8bit", "revision": "0754a2fa1ed5996497f0d738f0776fefbe65fa81", "url": "https://huggingface.co/lmstudio-community/granite-4.2-8b-MLX-8bit/tree/0754a2fa1ed5996497f0d738f0776fefbe65fa81", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 2, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-09-13", "title": "Reasoning support not detected by llama.cpp", "url": "https://huggingface.co/ibm-granite/granite-4.2-8b/discussions/3" } ], "notes": [] }, { "slug": "ling-3.0-tiny", "name": "Ling-3.0-tiny", "hf_repo": "inclusionAI/Ling-3.0-tiny", "hf_revision": "9a98e35fe1c9ee255f78dd64771c7ae15a799481", "source_url": "https://huggingface.co/inclusionAI/Ling-3.0-tiny", "org": "inclusionAI", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08", "precision": "month", "url": "https://artificialanalysis.ai/models/ling-3-0-tiny", "status": "verified_secondary_aa_release_metadata", "note": "AA explicitly reports release month; exact primary-source public release day remains unknown." }, "params": { "total_b": 7.9, "active_b": 1.3, "moe": true, "url": "https://huggingface.co/inclusionAI/Ling-3.0-tiny/blob/9a98e35fe1c9ee255f78dd64771c7ae15a799481/README.md", "active_url": "https://huggingface.co/inclusionAI/Ling-3.0-tiny/blob/9a98e35fe1c9ee255f78dd64771c7ae15a799481/README.md", "scope": "Current publisher card; active count changed in earlier card revisions." }, "arch_label": "MoE", "context": { "native": 131072, "url": "https://huggingface.co/inclusionAI/Ling-3.0-tiny/blob/9a98e35fe1c9ee255f78dd64771c7ae15a799481/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/inclusionAI/Ling-3.0-tiny/blob/9a98e35fe1c9ee255f78dd64771c7ae15a799481/LICENSE" }, "quality": { "rank": 57, "aa": { "value": 11.0636233769305, "variant": "Ling 3.0 Tiny", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/ling-3-0-tiny", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/inclusionAI/Ling-3.0-tiny/blob/9a98e35fe1c9ee255f78dd64771c7ae15a799481/config.json", "dtype": "FP16", "note": "Architecture needs backend-specific treatment: shared K/V or cross-layer cache, looped layers, MoVA, MLA/KDA or diffusion. No naive all-layers GQA estimate is supplied." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 4823894944, "repo": "inclusionAI/Ling-3.0-tiny-GGUF", "revision": "01b850e29ed5b44e2b810160ff980a9bb353296f", "url": "https://huggingface.co/inclusionAI/Ling-3.0-tiny-GGUF/tree/01b850e29ed5b44e2b810160ff980a9bb353296f", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 5635443616, "repo": "inclusionAI/Ling-3.0-tiny-GGUF", "revision": "01b850e29ed5b44e2b810160ff980a9bb353296f", "url": "https://huggingface.co/inclusionAI/Ling-3.0-tiny-GGUF/tree/01b850e29ed5b44e2b810160ff980a9bb353296f", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 8408187808, "repo": "inclusionAI/Ling-3.0-tiny-GGUF", "revision": "01b850e29ed5b44e2b810160ff980a9bb353296f", "url": "https://huggingface.co/inclusionAI/Ling-3.0-tiny-GGUF/tree/01b850e29ed5b44e2b810160ff980a9bb353296f", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 15803475168, "repo": "NANI-Nithin/Ling-3.0-tiny-GGUF", "revision": "dca7ccbeacbe284f084bc87f0e8fee128cd0e448", "url": "https://huggingface.co/NANI-Nithin/Ling-3.0-tiny-GGUF/tree/dca7ccbeacbe284f084bc87f0e8fee128cd0e448", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 4447478988, "repo": "rapid-mlx/Ling-3.0-tiny-MLX-4bit", "revision": "9ff45b05f51c42631b70f8d2f8e0d9fd4470edfe", "url": "https://huggingface.co/rapid-mlx/Ling-3.0-tiny-MLX-4bit/tree/9ff45b05f51c42631b70f8d2f8e0d9fd4470edfe", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "MLX_8bit" ], "community": [ { "platform": "Hugging Face", "date": "2026-08-30", "title": "Best model for its size, people are not using due to absense of GGUFs.", "url": "https://huggingface.co/inclusionAI/Ling-3.0-tiny/discussions/12" }, { "platform": "Hugging Face", "date": "2026-08-18", "title": "MTP support?", "url": "https://huggingface.co/inclusionAI/Ling-3.0-tiny/discussions/11" } ], "notes": [ "Exact public release day not verified from a primary announcement; repository creation and commit dates are deliberately not substituted." ] }, { "slug": "g9v3-3b", "name": "G9v3-3B", "hf_repo": "AI9Stars/G9v3-3B", "hf_revision": "d9553445ff92dbce667381954c9699fbcbc924f9", "source_url": "https://huggingface.co/AI9Stars/G9v3-3B", "org": "AI9Stars", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-07", "precision": "month", "url": "https://artificialanalysis.ai/models/g9v3-3b", "status": "verified_secondary_aa_release_metadata", "note": "AA explicitly reports release month; exact primary-source public release day remains unknown." }, "params": { "total_b": 3, "active_b": 3, "moe": false, "url": "https://huggingface.co/AI9Stars/G9v3-3B/blob/d9553445ff92dbce667381954c9699fbcbc924f9/README.md", "active_url": null, "scope": "Publisher approximate total." }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/AI9Stars/G9v3-3B/blob/d9553445ff92dbce667381954c9699fbcbc924f9/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/AI9Stars/G9v3-3B/blob/d9553445ff92dbce667381954c9699fbcbc924f9/LICENSE" }, "quality": { "rank": 58, "aa": { "value": 10.841600440564, "variant": "G9v3-3B", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/g9v3-3b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": 53248, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/AI9Stars/G9v3-3B/blob/d9553445ff92dbce667381954c9699fbcbc924f9/config.json", "dtype": "FP16", "note": "Ordinary GQA cache estimate; actual allocation and padding vary by backend." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 1843804896, "repo": "mradermacher/G9v3-3B-GGUF", "revision": "fe73312c4b4b13caf57bbc410b91478856cfd2cd", "url": "https://huggingface.co/mradermacher/G9v3-3B-GGUF/tree/fe73312c4b4b13caf57bbc410b91478856cfd2cd", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 2141338336, "repo": "mradermacher/G9v3-3B-GGUF", "revision": "fe73312c4b4b13caf57bbc410b91478856cfd2cd", "url": "https://huggingface.co/mradermacher/G9v3-3B-GGUF/tree/fe73312c4b4b13caf57bbc410b91478856cfd2cd", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 3181230816, "repo": "mradermacher/G9v3-3B-GGUF", "revision": "fe73312c4b4b13caf57bbc410b91478856cfd2cd", "url": "https://huggingface.co/mradermacher/G9v3-3B-GGUF/tree/fe73312c4b4b13caf57bbc410b91478856cfd2cd", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 5982894816, "repo": "mradermacher/G9v3-3B-GGUF", "revision": "fe73312c4b4b13caf57bbc410b91478856cfd2cd", "url": "https://huggingface.co/mradermacher/G9v3-3B-GGUF/tree/fe73312c4b4b13caf57bbc410b91478856cfd2cd", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 1681564038, "repo": "cof139/G9v3-3B-mlx-4Bit", "revision": "076ed58eed5a29dc7a27cf16d184db59550f09cb", "url": "https://huggingface.co/cof139/G9v3-3B-mlx-4Bit/tree/076ed58eed5a29dc7a27cf16d184db59550f09cb", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "MLX_8bit" ], "community": [], "notes": [ "Exact public release day not verified from a primary announcement; repository creation and commit dates are deliberately not substituted." ] }, { "slug": "trinity-large-thinking", "name": "Trinity-Large-Thinking", "hf_repo": "arcee-ai/Trinity-Large-Thinking", "hf_revision": "dc6a99b6b74202880e31491a445a94884c97890c", "source_url": "https://huggingface.co/arcee-ai/Trinity-Large-Thinking", "org": "arcee-ai", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-04-01", "precision": "day", "url": "https://www.arcee.ai/blog/trinity-large-thinking", "status": "verified_primary", "note": "Official blog explicitly says weights available today." }, "params": { "total_b": 398, "active_b": 13, "moe": true, "url": "https://huggingface.co/arcee-ai/Trinity-Large-Thinking/raw/dc6a99b6b74202880e31491a445a94884c97890c/README.md", "active_url": "https://huggingface.co/arcee-ai/Trinity-Large-Thinking/raw/dc6a99b6b74202880e31491a445a94884c97890c/README.md", "scope": null }, "arch_label": "MoE · sliding-window attn", "context": { "native": 262144, "url": "https://huggingface.co/arcee-ai/Trinity-Large-Thinking/raw/dc6a99b6b74202880e31491a445a94884c97890c/config.json", "extended": null }, "license": { "value": "openmdw-1.1", "url": "https://huggingface.co/arcee-ai/Trinity-Large-Thinking/blob/dc6a99b6b74202880e31491a445a94884c97890c/LICENSE" }, "quality": { "rank": 59, "aa": { "value": 10.8242425924218, "variant": "Trinity Large Thinking", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/trinity-large-thinking", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": 61440, "win_per_token": null, "win_tokens": null, "win_max": 754974720, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/arcee-ai/Trinity-Large-Thinking/raw/dc6a99b6b74202880e31491a445a94884c97890c/config.json", "dtype": "BF16", "note": "BF16/FP16 KV estimate for full-attention layers only, batch=1. Sliding-window terms and fixed recurrent state are separate; excludes alignment, graphs, buffers, vision and MTP/drafters." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 241848238432, "repo": "bartowski/arcee-ai_Trinity-Large-Thinking-GGUF", "revision": "3854f407a7e4f2c025f8d55b9ed34fdc5970d085", "url": "https://huggingface.co/bartowski/arcee-ai_Trinity-Large-Thinking-GGUF/tree/3854f407a7e4f2c025f8d55b9ed34fdc5970d085", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 7, "status": "measured_file_manifest" }, "GGUF_Q5_K_M": { "bytes": 283607089568, "repo": "bartowski/arcee-ai_Trinity-Large-Thinking-GGUF", "revision": "3854f407a7e4f2c025f8d55b9ed34fdc5970d085", "url": "https://huggingface.co/bartowski/arcee-ai_Trinity-Large-Thinking-GGUF/tree/3854f407a7e4f2c025f8d55b9ed34fdc5970d085", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 8, "status": "measured_file_manifest" }, "GGUF_Q8_0": { "bytes": 423684403840, "repo": "bartowski/arcee-ai_Trinity-Large-Thinking-GGUF", "revision": "3854f407a7e4f2c025f8d55b9ed34fdc5970d085", "url": "https://huggingface.co/bartowski/arcee-ai_Trinity-Large-Thinking-GGUF/tree/3854f407a7e4f2c025f8d55b9ed34fdc5970d085", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 11, "status": "measured_file_manifest" }, "MLX_4bit": { "bytes": 224255014880, "repo": "mlx-community/Trinity-Large-Thinking-4bit", "revision": "2c7f70cbb5a3ab04b7a5e9b5e9274dc1e717b0ae", "url": "https://huggingface.co/mlx-community/Trinity-Large-Thinking-4bit/tree/2c7f70cbb5a3ab04b7a5e9b5e9274dc1e717b0ae", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 47, "status": "measured_file_manifest", "mixed": true }, "MLX_8bit": { "bytes": 423551038416, "repo": "mlx-community/Trinity-Large-Thinking-8bit", "revision": "332efb64168a091e22150c55532454eaf0a17568", "url": "https://huggingface.co/mlx-community/Trinity-Large-Thinking-8bit/tree/332efb64168a091e22150c55532454eaf0a17568", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 82, "status": "measured_file_manifest", "mixed": true } }, "artifacts_missing": [ "GGUF_FP16", "GGUF_BF16" ], "community": [], "notes": [] }, { "slug": "nemotron-3-nano-omni-30b-a3b-reasoning", "name": "Nemotron 3 Nano Omni 30B A3B Reasoning", "hf_repo": "nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16", "hf_revision": "e5e9932441de940c9a62185c870ea5bcd4cd24e2", "source_url": "https://huggingface.co/nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16", "org": "nvidia", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-04-28", "precision": "day", "url": "https://huggingface.co/nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16/blob/e5e9932441de940c9a62185c870ea5bcd4cd24e2/README.md", "status": "verified_primary", "note": "Model card explicit Release Date." }, "params": { "total_b": 30, "active_b": 3, "moe": true, "url": "https://huggingface.co/nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16/blob/e5e9932441de940c9a62185c870ea5bcd4cd24e2/README.md", "active_url": "https://huggingface.co/nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16/blob/e5e9932441de940c9a62185c870ea5bcd4cd24e2/README.md", "scope": "Language-backbone nominal count; multimodal encoders add weights." }, "arch_label": "MoE · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16/blob/e5e9932441de940c9a62185c870ea5bcd4cd24e2/config.json", "extended": null }, "license": { "value": "nvidia-open-model-agreement", "url": "https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-agreement/" }, "quality": { "rank": 60, "aa": { "value": 10.2506761710987, "variant": "Nemotron 3 Nano Omni 30B A3B Reasoning", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/nemotron-3-nano-omni-30b-a3b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16/blob/e5e9932441de940c9a62185c870ea5bcd4cd24e2/config.json", "dtype": "FP16", "note": "Architecture needs backend-specific treatment: shared K/V or cross-layer cache, looped layers, MoVA, MLA/KDA or diffusion. No naive all-layers GQA estimate is supplied." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 24515129536, "repo": "lmstudio-community/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-GGUF", "revision": "49a5b1ae335e30e775d4eb03f822b52a63ea4271", "url": "https://huggingface.co/lmstudio-community/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-GGUF/tree/49a5b1ae335e30e775d4eb03f822b52a63ea4271", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16.gguf", "bytes": 1589506176, "url": "https://huggingface.co/lmstudio-community/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-GGUF/blob/49a5b1ae335e30e775d4eb03f822b52a63ea4271/mmproj-Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 33585499584, "repo": "unsloth/NVIDIA-Nemotron-3-Nano-Omni-30B-A3B-Reasoning-GGUF", "revision": "571758804835f56154718683f5c0e388b7d0fef9", "url": "https://huggingface.co/unsloth/NVIDIA-Nemotron-3-Nano-Omni-30B-A3B-Reasoning-GGUF/tree/571758804835f56154718683f5c0e388b7d0fef9", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 1589506304, "url": "https://huggingface.co/unsloth/NVIDIA-Nemotron-3-Nano-Omni-30B-A3B-Reasoning-GGUF/blob/571758804835f56154718683f5c0e388b7d0fef9/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 1587540224, "url": "https://huggingface.co/unsloth/NVIDIA-Nemotron-3-Nano-Omni-30B-A3B-Reasoning-GGUF/blob/571758804835f56154718683f5c0e388b7d0fef9/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 3167613184, "url": "https://huggingface.co/unsloth/NVIDIA-Nemotron-3-Nano-Omni-30B-A3B-Reasoning-GGUF/blob/571758804835f56154718683f5c0e388b7d0fef9/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 19633498265, "repo": "mlx-community/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-4bit", "revision": "736a5aca22c2183bfb04534739fc8a6cd8422810", "url": "https://huggingface.co/mlx-community/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-4bit/tree/736a5aca22c2183bfb04534739fc8a6cd8422810", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 4, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 35771930827, "repo": "mlx-community/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-8bit", "revision": "8a32c7a51910059f86062fc947261de07cc9be98", "url": "https://huggingface.co/mlx-community/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-8bit/tree/8a32c7a51910059f86062fc947261de07cc9be98", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 8, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_Q5_K_M", "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-06-21", "title": "Please support MTP", "url": "https://huggingface.co/nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16/discussions/13" }, { "platform": "Hugging Face", "date": "2026-05-13", "title": "`modeling_nemotron_h.py` missing `supports_gradient_checkpointing = True` on `NemotronHPreTrainedModel`", "url": "https://huggingface.co/nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16/discussions/12" } ], "notes": [] }, { "slug": "north-mini-code", "name": "North Mini Code", "hf_repo": "CohereLabs/North-Mini-Code-1.0", "hf_revision": "d11e61a842617a22dc328552fa5bb86231ee4f37", "source_url": "https://huggingface.co/CohereLabs/North-Mini-Code-1.0", "org": "CohereLabs", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-06-09", "precision": "day", "url": "https://docs.cohere.com/v2/changelog", "status": "verified_primary", "note": "Official dated release changelog states open weights available." }, "params": { "total_b": 30, "active_b": 3, "moe": true, "url": "https://huggingface.co/CohereLabs/North-Mini-Code-1.0/blob/d11e61a842617a22dc328552fa5bb86231ee4f37/README.md", "active_url": "https://huggingface.co/CohereLabs/North-Mini-Code-1.0/blob/d11e61a842617a22dc328552fa5bb86231ee4f37/README.md", "scope": "Model-card nominal active count; later Cohere megakernel post says 3.3B." }, "arch_label": "MoE · sliding-window attn", "context": { "native": 256000, "url": "https://huggingface.co/CohereLabs/North-Mini-Code-1.0/blob/d11e61a842617a22dc328552fa5bb86231ee4f37/README.md", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/CohereLabs/North-Mini-Code-1.0/blob/d11e61a842617a22dc328552fa5bb86231ee4f37/LICENSE" }, "quality": { "rank": 61, "aa": { "value": 9.91462776590372, "variant": "North Mini Code", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/north-mini-code", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": 26624, "win_per_token": 73728, "win_tokens": 4096, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/CohereLabs/North-Mini-Code-1.0/blob/d11e61a842617a22dc328552fa5bb86231ee4f37/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 18744024640, "repo": "bartowski/North-Mini-Code-1.0-GGUF", "revision": "6ff6563002170723a6f7a672bf4c99775be6c0dd", "url": "https://huggingface.co/bartowski/North-Mini-Code-1.0-GGUF/tree/6ff6563002170723a6f7a672bf4c99775be6c0dd", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 21845253696, "repo": "bartowski/North-Mini-Code-1.0-GGUF", "revision": "6ff6563002170723a6f7a672bf4c99775be6c0dd", "url": "https://huggingface.co/bartowski/North-Mini-Code-1.0-GGUF/tree/6ff6563002170723a6f7a672bf4c99775be6c0dd", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 32437263936, "repo": "bartowski/North-Mini-Code-1.0-GGUF", "revision": "6ff6563002170723a6f7a672bf4c99775be6c0dd", "url": "https://huggingface.co/bartowski/North-Mini-Code-1.0-GGUF/tree/6ff6563002170723a6f7a672bf4c99775be6c0dd", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 18495219875, "repo": "mlx-community/North-Mini-Code-1.0-4bit", "revision": "dfbe084dfa26e241345af99ca32848f38fd865f9", "url": "https://huggingface.co/mlx-community/North-Mini-Code-1.0-4bit/tree/dfbe084dfa26e241345af99ca32848f38fd865f9", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 4, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 33268607199, "repo": "aaronmwelborn/North-Mini-Code-1.0-8bit", "revision": "26e3f034fef0c48e6b93d43cdb3156fdedef7a97", "url": "https://huggingface.co/aaronmwelborn/North-Mini-Code-1.0-8bit/tree/26e3f034fef0c48e6b93d43cdb3156fdedef7a97", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 7, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-06-20", "title": "MLX chat_template.jinja", "url": "https://huggingface.co/CohereLabs/North-Mini-Code-1.0/discussions/12" }, { "platform": "Hugging Face", "date": "2026-06-09", "title": "llama.cpp?", "url": "https://huggingface.co/CohereLabs/North-Mini-Code-1.0/discussions/4" } ], "notes": [] }, { "slug": "diffusiongemma-26b-a4b", "name": "DiffusionGemma 26B A4B", "hf_repo": "google/diffusiongemma-26B-A4B-it", "hf_revision": "f7f5b7f5fa82ffc52addd066915886d497f5517b", "source_url": "https://huggingface.co/google/diffusiongemma-26B-A4B-it", "org": "google", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-06-10", "precision": "day", "url": "https://blog.google/innovation-and-ai/technology/developers-tools/diffusion-gemma-faster-text-generation/", "status": "verified_primary", "note": "Publisher announcement." }, "params": { "total_b": 25.2, "active_b": 3.8, "moe": true, "url": "https://huggingface.co/google/diffusiongemma-26B-A4B-it/blob/f7f5b7f5fa82ffc52addd066915886d497f5517b/README.md", "active_url": "https://huggingface.co/google/diffusiongemma-26B-A4B-it/blob/f7f5b7f5fa82ffc52addd066915886d497f5517b/README.md", "scope": "Decoder count plus vision/diffusion components; not ordinary autoregressive generation." }, "arch_label": "MoE · sliding-window attn", "context": { "native": 262144, "url": "https://huggingface.co/google/diffusiongemma-26B-A4B-it/blob/f7f5b7f5fa82ffc52addd066915886d497f5517b/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://ai.google.dev/gemma/docs/gemma_4_license" }, "quality": { "rank": 62, "aa": { "value": 9.51014305662543, "variant": "DiffusionGemma 26B A4B", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/diffusiongemma-26b-a4b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/google/diffusiongemma-26B-A4B-it/blob/f7f5b7f5fa82ffc52addd066915886d497f5517b/config.json", "dtype": "FP16", "note": "Architecture needs backend-specific treatment: shared K/V or cross-layer cache, looped layers, MoVA, MLA/KDA or diffusion. No naive all-layers GQA estimate is supplied." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 16806810208, "repo": "unsloth/diffusiongemma-26B-A4B-it-GGUF", "revision": "f4183a2c7a354128d02545752303c4354d165bf0", "url": "https://huggingface.co/unsloth/diffusiongemma-26B-A4B-it-GGUF/tree/f4183a2c7a354128d02545752303c4354d165bf0", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 19145543264, "repo": "unsloth/diffusiongemma-26B-A4B-it-GGUF", "revision": "f4183a2c7a354128d02545752303c4354d165bf0", "url": "https://huggingface.co/unsloth/diffusiongemma-26B-A4B-it-GGUF/tree/f4183a2c7a354128d02545752303c4354d165bf0", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 26878831200, "repo": "unsloth/diffusiongemma-26B-A4B-it-GGUF", "revision": "f4183a2c7a354128d02545752303c4354d165bf0", "url": "https://huggingface.co/unsloth/diffusiongemma-26B-A4B-it-GGUF/tree/f4183a2c7a354128d02545752303c4354d165bf0", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 16543055405, "repo": "mlx-community/diffusiongemma-26B-A4B-it-4bit", "revision": "a7a81407613811e8ba63af92ac0d852b809e191f", "url": "https://huggingface.co/mlx-community/diffusiongemma-26B-A4B-it-4bit/tree/a7a81407613811e8ba63af92ac0d852b809e191f", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 4, "status": "verified_hf_file_sizes", "mixed": true, "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 27972591174, "repo": "mlx-community/diffusiongemma-26B-A4B-it-8bit", "revision": "7b95e3887078ba56283c24f2578d6e5a06b9d7e8", "url": "https://huggingface.co/mlx-community/diffusiongemma-26B-A4B-it-8bit/tree/7b95e3887078ba56283c24f2578d6e5a06b9d7e8", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 6, "status": "verified_hf_file_sizes", "mixed": true, "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-06-12", "title": "ADD FAST-dLLM mode for more speed", "url": "https://huggingface.co/google/diffusiongemma-26B-A4B-it/discussions/9" }, { "platform": "Hugging Face", "date": "2026-06-11", "title": "RuntimeError: Tensor.item() cannot be called on meta tensors", "url": "https://huggingface.co/google/diffusiongemma-26B-A4B-it/discussions/6" } ], "notes": [] }, { "slug": "qwen3-coder-next", "name": "Qwen3-Coder-Next", "hf_repo": "Qwen/Qwen3-Coder-Next", "hf_revision": "a7fbcb5c0e12d62a448eaa0e260346bf5dcc0feb", "source_url": "https://huggingface.co/Qwen/Qwen3-Coder-Next", "org": "Qwen", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-02-02", "precision": "day", "url": "https://qwen.ai/blog?_bhlid=9abbd8a3cb1b0f1859b6c9f5bb81f296904c7e02&id=qwen3-coder-next", "status": "verified_primary", "note": "Indexed official blog explicitly dated Feb2; widespread community/HF release coverage Feb3. Preserve date discrepancy; use publisher date." }, "params": { "total_b": 80, "active_b": 3, "moe": true, "url": "https://huggingface.co/Qwen/Qwen3-Coder-Next/raw/a7fbcb5c0e12d62a448eaa0e260346bf5dcc0feb/README.md", "active_url": "https://huggingface.co/Qwen/Qwen3-Coder-Next/raw/a7fbcb5c0e12d62a448eaa0e260346bf5dcc0feb/README.md", "scope": null }, "arch_label": "MoE · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/Qwen/Qwen3-Coder-Next/raw/a7fbcb5c0e12d62a448eaa0e260346bf5dcc0feb/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/Qwen/Qwen3-Coder-Next/blob/main/LICENSE" }, "quality": { "rank": 63, "aa": { "value": 9.23682203219174, "variant": "Qwen3 Coder Next", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/qwen3-coder-next", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": 24576, "win_per_token": null, "win_tokens": null, "win_max": 0, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3-Coder-Next/raw/a7fbcb5c0e12d62a448eaa0e260346bf5dcc0feb/config.json", "dtype": "BF16", "note": "BF16/FP16 KV estimate for full-attention layers only, batch=1. Sliding-window terms and fixed recurrent state are separate; excludes alignment, graphs, buffers, vision and MTP/drafters." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 48528320544, "repo": "unsloth/Qwen3-Coder-Next-GGUF", "revision": "ce09c67b53bc8739eef83fe67b2f5d293c270632", "url": "https://huggingface.co/unsloth/Qwen3-Coder-Next-GGUF/tree/ce09c67b53bc8739eef83fe67b2f5d293c270632", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 1, "status": "measured_file_manifest" }, "GGUF_Q5_K_M": { "bytes": 56847886688, "repo": "unsloth/Qwen3-Coder-Next-GGUF", "revision": "ce09c67b53bc8739eef83fe67b2f5d293c270632", "url": "https://huggingface.co/unsloth/Qwen3-Coder-Next-GGUF/tree/ce09c67b53bc8739eef83fe67b2f5d293c270632", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 3, "status": "measured_file_manifest" }, "GGUF_Q8_0": { "bytes": 84812255584, "repo": "unsloth/Qwen3-Coder-Next-GGUF", "revision": "ce09c67b53bc8739eef83fe67b2f5d293c270632", "url": "https://huggingface.co/unsloth/Qwen3-Coder-Next-GGUF/tree/ce09c67b53bc8739eef83fe67b2f5d293c270632", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 3, "status": "measured_file_manifest" }, "GGUF_BF16": { "bytes": 159458415168, "repo": "unsloth/Qwen3-Coder-Next-GGUF", "revision": "ce09c67b53bc8739eef83fe67b2f5d293c270632", "url": "https://huggingface.co/unsloth/Qwen3-Coder-Next-GGUF/tree/ce09c67b53bc8739eef83fe67b2f5d293c270632", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 4, "status": "measured_file_manifest" }, "MLX_4bit": { "bytes": 44844286500, "repo": "mlx-community/Qwen3-Coder-Next-4bit", "revision": "7b9321eabb85ce79625cac3f61ea691e4ea984b5", "url": "https://huggingface.co/mlx-community/Qwen3-Coder-Next-4bit/tree/7b9321eabb85ce79625cac3f61ea691e4ea984b5", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 9, "status": "measured_file_manifest", "mixed": true }, "MLX_8bit": { "bytes": 84655572377, "repo": "mlx-community/Qwen3-Coder-Next-8bit", "revision": "6d3c664dc8539a711783391484fd6784c51fd8fa", "url": "https://huggingface.co/mlx-community/Qwen3-Coder-Next-8bit/tree/6d3c664dc8539a711783391484fd6784c51fd8fa", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 17, "status": "measured_file_manifest", "mixed": true } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [], "notes": [] }, { "slug": "granite-4.2-3b", "name": "Granite 4.2 3B", "hf_repo": "ibm-granite/granite-4.2-3b", "hf_revision": "e459acceac81e5fe67c07d9cfc72329a332e7eb1", "source_url": "https://huggingface.co/ibm-granite/granite-4.2-3b", "org": "ibm-granite", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08-25", "precision": "day", "url": "https://huggingface.co/ibm-granite/granite-4.2-3b/blob/e459acceac81e5fe67c07d9cfc72329a332e7eb1/README.md", "status": "verified_primary", "note": "Model card explicit Release Date." }, "params": { "total_b": 3, "active_b": 3, "moe": false, "url": "https://huggingface.co/ibm-granite/granite-4.2-3b/blob/e459acceac81e5fe67c07d9cfc72329a332e7eb1/README.md", "active_url": null, "scope": "Publisher nominal size." }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/ibm-granite/granite-4.2-3b/blob/e459acceac81e5fe67c07d9cfc72329a332e7eb1/config.json", "extended": 524288 }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/ibm-granite/granite-4.2-3b/blob/e459acceac81e5fe67c07d9cfc72329a332e7eb1/LICENSE" }, "quality": { "rank": 64, "aa": { "value": 9.05588232376756, "variant": "Granite 4.2 3B", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/granite-4-2-3b", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1292.9351611166, "lower": 1281.4396914316, "upper": 1304.4306308015, "votes": 3251, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": 81920, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/ibm-granite/granite-4.2-3b/blob/e459acceac81e5fe67c07d9cfc72329a332e7eb1/config.json", "dtype": "FP16", "note": "Ordinary GQA cache estimate; actual allocation and padding vary by backend." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 2244011552, "repo": "ibm-granite/granite-4.2-3b-GGUF", "revision": "c40945d71cd90f249a56985e8155551a9188dc30", "url": "https://huggingface.co/ibm-granite/granite-4.2-3b-GGUF/tree/c40945d71cd90f249a56985e8155551a9188dc30", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 2613634592, "repo": "ibm-granite/granite-4.2-3b-GGUF", "revision": "c40945d71cd90f249a56985e8155551a9188dc30", "url": "https://huggingface.co/ibm-granite/granite-4.2-3b-GGUF/tree/c40945d71cd90f249a56985e8155551a9188dc30", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 3892651552, "repo": "ibm-granite/granite-4.2-3b-GGUF", "revision": "c40945d71cd90f249a56985e8155551a9188dc30", "url": "https://huggingface.co/ibm-granite/granite-4.2-3b-GGUF/tree/c40945d71cd90f249a56985e8155551a9188dc30", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 7323462208, "repo": "mradermacher/granite-4.2-3b-GGUF", "revision": "a9c2f30cb406857bcafe3169a4388a2292414aa3", "url": "https://huggingface.co/mradermacher/granite-4.2-3b-GGUF/tree/a9c2f30cb406857bcafe3169a4388a2292414aa3", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 2059005430, "repo": "lmstudio-community/granite-4.2-3b-MLX-4bit", "revision": "4fe06f1699e99e288bf765ebc57f629cbca3bbfe", "url": "https://huggingface.co/lmstudio-community/granite-4.2-3b-MLX-4bit/tree/4fe06f1699e99e288bf765ebc57f629cbca3bbfe", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 3888771048, "repo": "lmstudio-community/granite-4.2-3b-MLX-8bit", "revision": "7c630b526561db6b4641767c54595fe494176695", "url": "https://huggingface.co/lmstudio-community/granite-4.2-3b-MLX-8bit/tree/7c630b526561db6b4641767c54595fe494176695", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-09-10", "title": "Clarification on enabling and evaluating 512K context", "url": "https://huggingface.co/ibm-granite/granite-4.2-3b/discussions/3" } ], "notes": [] }, { "slug": "tri-21b-think", "name": "Tri-21B-Think", "hf_repo": "trillionlabs/Tri-21B-Think", "hf_revision": "301724769f88cd71f0207f44d021178217a77ad6", "source_url": "https://huggingface.co/trillionlabs/Tri-21B-Think", "org": "trillionlabs", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-02", "precision": "month", "url": "https://artificialanalysis.ai/models/tri-21b-think-v0-5", "status": "verified_secondary_aa_release_metadata", "note": "AA explicitly reports release month; exact primary-source public release day remains unknown." }, "params": { "total_b": 20.73, "active_b": 20.73, "moe": false, "url": "https://huggingface.co/trillionlabs/Tri-21B-Think/blob/301724769f88cd71f0207f44d021178217a77ad6/README.md", "active_url": null, "scope": "Publisher model-card count." }, "arch_label": "Dense", "context": { "native": 32768, "url": "https://huggingface.co/trillionlabs/Tri-21B-Think/blob/301724769f88cd71f0207f44d021178217a77ad6/config.json", "extended": 262144 }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/trillionlabs/Tri-21B-Think/blob/301724769f88cd71f0207f44d021178217a77ad6/LICENSE" }, "quality": { "rank": 65, "aa": { "value": 8.98704047108094, "variant": "Tri-21B-Think", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/tri-21b-think-v0-5", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": 204800, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/trillionlabs/Tri-21B-Think/blob/301724769f88cd71f0207f44d021178217a77ad6/config.json", "dtype": "FP16", "note": "Ordinary GQA cache estimate; actual allocation and padding vary by backend." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 12588064832, "repo": "mykor/Tri-21B-Think-GGUF", "revision": "8574fbbacb44d0315540e2268a7b1c329de13104", "url": "https://huggingface.co/mykor/Tri-21B-Think-GGUF/tree/8574fbbacb44d0315540e2268a7b1c329de13104", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 14732075072, "repo": "mykor/Tri-21B-Think-GGUF", "revision": "8574fbbacb44d0315540e2268a7b1c329de13104", "url": "https://huggingface.co/mykor/Tri-21B-Think-GGUF/tree/8574fbbacb44d0315540e2268a7b1c329de13104", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 22029447232, "repo": "mykor/Tri-21B-Think-GGUF", "revision": "8574fbbacb44d0315540e2268a7b1c329de13104", "url": "https://huggingface.co/mykor/Tri-21B-Think-GGUF/tree/8574fbbacb44d0315540e2268a7b1c329de13104", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 41459232864, "repo": "KnutJaegersberg/Tri-21B-Think-gguf", "revision": "426c3d57763fc2e8bf8b2cd3753fcb12e033f199", "url": "https://huggingface.co/KnutJaegersberg/Tri-21B-Think-gguf/tree/426c3d57763fc2e8bf8b2cd3753fcb12e033f199", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "MLX_4bit", "MLX_8bit" ], "community": [ { "platform": "Hugging Face", "date": "2026-02-19", "title": "Device assertion errors with transformers", "url": "https://huggingface.co/trillionlabs/Tri-21B-Think/discussions/1" }, { "platform": "Hugging Face", "date": "2026-03-13", "title": "Installation Video and Testing - Step by Step", "url": "https://huggingface.co/trillionlabs/Tri-21B-Think/discussions/2" } ], "notes": [ "Exact public release day not verified from a primary announcement; repository creation and commit dates are deliberately not substituted." ] }, { "slug": "gemma-4-e4b", "name": "Gemma 4 E4B (it)", "hf_repo": "google/gemma-4-E4B-it", "hf_revision": "ee0ef6023621cff504d758262d4e04895a5af4a2", "source_url": "https://huggingface.co/google/gemma-4-E4B-it", "org": "google", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-04-02", "precision": "day", "url": "https://blog.google/innovation-and-ai/technology/developers-tools/gemma-4/", "status": "verified_primary", "note": "Public launch announcement; distinguish earlier repository creation." }, "params": { "total_b": 8, "active_b": 8, "moe": false, "url": "https://huggingface.co/google/gemma-4-E4B-it/blob/ee0ef6023621cff504d758262d4e04895a5af4a2/README.md", "active_url": null, "scope": "4.5B effective; PLE embeddings make total 8B. Not a 4B stored-weight model." }, "arch_label": "Dense · sliding-window attn", "context": { "native": 131072, "url": "https://huggingface.co/google/gemma-4-E4B-it/blob/ee0ef6023621cff504d758262d4e04895a5af4a2/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://ai.google.dev/gemma/docs/gemma_4_license" }, "quality": { "rank": 66, "aa": { "value": 8.90908441839731, "variant": "Gemma 4 E4B (Reasoning)", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/gemma-4-e4b", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/google/gemma-4-E4B-it/blob/ee0ef6023621cff504d758262d4e04895a5af4a2/config.json", "dtype": "FP16", "note": "Architecture needs backend-specific treatment: shared K/V or cross-layer cache, looped layers, MoVA, MLA/KDA or diffusion. No naive all-layers GQA estimate is supplied." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 4977171584, "repo": "unsloth/gemma-4-E4B-it-GGUF", "revision": "bfc15c382204943c3a8fff0c750b94ae2364d7a3", "url": "https://huggingface.co/unsloth/gemma-4-E4B-it-GGUF/tree/bfc15c382204943c3a8fff0c750b94ae2364d7a3", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 991552320, "url": "https://huggingface.co/unsloth/gemma-4-E4B-it-GGUF/blob/bfc15c382204943c3a8fff0c750b94ae2364d7a3/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 990372672, "url": "https://huggingface.co/unsloth/gemma-4-E4B-it-GGUF/blob/bfc15c382204943c3a8fff0c750b94ae2364d7a3/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1912464192, "url": "https://huggingface.co/unsloth/gemma-4-E4B-it-GGUF/blob/bfc15c382204943c3a8fff0c750b94ae2364d7a3/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 5481798784, "repo": "unsloth/gemma-4-E4B-it-GGUF", "revision": "bfc15c382204943c3a8fff0c750b94ae2364d7a3", "url": "https://huggingface.co/unsloth/gemma-4-E4B-it-GGUF/tree/bfc15c382204943c3a8fff0c750b94ae2364d7a3", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 991552320, "url": "https://huggingface.co/unsloth/gemma-4-E4B-it-GGUF/blob/bfc15c382204943c3a8fff0c750b94ae2364d7a3/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 990372672, "url": "https://huggingface.co/unsloth/gemma-4-E4B-it-GGUF/blob/bfc15c382204943c3a8fff0c750b94ae2364d7a3/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1912464192, "url": "https://huggingface.co/unsloth/gemma-4-E4B-it-GGUF/blob/bfc15c382204943c3a8fff0c750b94ae2364d7a3/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 8192953472, "repo": "unsloth/gemma-4-E4B-it-GGUF", "revision": "bfc15c382204943c3a8fff0c750b94ae2364d7a3", "url": "https://huggingface.co/unsloth/gemma-4-E4B-it-GGUF/tree/bfc15c382204943c3a8fff0c750b94ae2364d7a3", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 991552320, "url": "https://huggingface.co/unsloth/gemma-4-E4B-it-GGUF/blob/bfc15c382204943c3a8fff0c750b94ae2364d7a3/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 990372672, "url": "https://huggingface.co/unsloth/gemma-4-E4B-it-GGUF/blob/bfc15c382204943c3a8fff0c750b94ae2364d7a3/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1912464192, "url": "https://huggingface.co/unsloth/gemma-4-E4B-it-GGUF/blob/bfc15c382204943c3a8fff0c750b94ae2364d7a3/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 5146800534, "repo": "mlx-community/gemma-4-E4B-it-4bit", "revision": "475b9088d29754a3379866cf5aeb6b41acd313c2", "url": "https://huggingface.co/mlx-community/gemma-4-E4B-it-4bit/tree/475b9088d29754a3379866cf5aeb6b41acd313c2", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 8880976964, "repo": "mlx-community/gemma-4-E4B-it-8bit", "revision": "4255b21bd9a9d3fc807ef7abd80373f5e3a52a73", "url": "https://huggingface.co/mlx-community/gemma-4-E4B-it-8bit/tree/4255b21bd9a9d3fc807ef7abd80373f5e3a52a73", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 2, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-10-10", "title": "A DSpark draft model for Gemma-4-E4B-it (up to 4.7x speedup; higher acceptance than MTP on 4/9 benchmarks)", "url": "https://huggingface.co/google/gemma-4-E4B-it/discussions/51" }, { "platform": "Hugging Face", "date": "2026-09-21", "title": "[Exploration] Applying Native Dual-Regime & KV-Cache Handoff to this model variant?", "url": "https://huggingface.co/google/gemma-4-E4B-it/discussions/48" }, { "platform": "Reddit", "date": "2026-10-16", "title": null, "url": "https://www.reddit.com/r/LocalLLaMA/comments/1x7g8cc/why_do_we_have_models_under_15b_so_rarely_anymore/" } ], "notes": [] }, { "slug": "minicpm5-1b", "name": "MiniCPM5-1B", "hf_repo": "openbmb/MiniCPM5-1B", "hf_revision": "87179e5c1f455ef22e6223592d2d61351b525bfc", "source_url": "https://huggingface.co/openbmb/MiniCPM5-1B", "org": "openbmb", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-05-19", "precision": "day", "url": "https://github.com/OpenBMB/MiniCPM", "status": "verified_primary", "note": "Official release changelog." }, "params": { "total_b": 1.080632832, "active_b": 1.080632832, "moe": false, "url": "https://huggingface.co/openbmb/MiniCPM5-1B/blob/87179e5c1f455ef22e6223592d2d61351b525bfc/README.md", "active_url": null, "scope": "Publisher exact total; non-embedding count 679,552,512." }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/openbmb/MiniCPM5-1B/blob/87179e5c1f455ef22e6223592d2d61351b525bfc/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/openbmb/MiniCPM5-1B/blob/87179e5c1f455ef22e6223592d2d61351b525bfc/LICENSE" }, "quality": { "rank": 67, "aa": { "value": 8.79664321370212, "variant": "MiniCPM5-1B (Reasoning)", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/minicpm5-1b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": 24576, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/openbmb/MiniCPM5-1B/blob/87179e5c1f455ef22e6223592d2d61351b525bfc/config.json", "dtype": "FP16", "note": "Ordinary GQA cache estimate; actual allocation and padding vary by backend." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 688065920, "repo": "openbmb/MiniCPM5-1B-GGUF", "revision": "3d55fac80935ae6456986ad2384b5cbcc4d6c948", "url": "https://huggingface.co/openbmb/MiniCPM5-1B-GGUF/tree/3d55fac80935ae6456986ad2384b5cbcc4d6c948", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 786862688, "repo": "Abiray/MiniCPM5-1B-GGUF", "revision": "0a15cbb4881b4282242bad0ff2630285ceec6855", "url": "https://huggingface.co/Abiray/MiniCPM5-1B-GGUF/tree/0a15cbb4881b4282242bad0ff2630285ceec6855", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 1153529216, "repo": "openbmb/MiniCPM5-1B-GGUF", "revision": "3d55fac80935ae6456986ad2384b5cbcc4d6c948", "url": "https://huggingface.co/openbmb/MiniCPM5-1B-GGUF/tree/3d55fac80935ae6456986ad2384b5cbcc4d6c948", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 2166551936, "repo": "openbmb/MiniCPM5-1B-GGUF", "revision": "3d55fac80935ae6456986ad2384b5cbcc4d6c948", "url": "https://huggingface.co/openbmb/MiniCPM5-1B-GGUF/tree/3d55fac80935ae6456986ad2384b5cbcc4d6c948", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 608026621, "repo": "openbmb/MiniCPM5-1B-MLX", "revision": "9879b18bf2928355fcdf4287635388a3665a40cb", "url": "https://huggingface.co/openbmb/MiniCPM5-1B-MLX/tree/9879b18bf2928355fcdf4287635388a3665a40cb", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 1148305460, "repo": "mlx-community/MiniCPM5-1B-8bit", "revision": "3d164befdfbe496a9e280704c42bcb34ebde9443", "url": "https://huggingface.co/mlx-community/MiniCPM5-1B-8bit/tree/3d164befdfbe496a9e280704c42bcb34ebde9443", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-07-26", "title": "Interpretation tools: SAE weights (english only) brain atlas SQLite database", "url": "https://huggingface.co/openbmb/MiniCPM5-1B/discussions/24" }, { "platform": "Hugging Face", "date": "2026-06-01", "title": "Any plan to support ltertlm?", "url": "https://huggingface.co/openbmb/MiniCPM5-1B/discussions/16" } ], "notes": [] }, { "slug": "sarvam-105b", "name": "Sarvam 105B", "hf_repo": "sarvamai/sarvam-105b", "hf_revision": "abe264bdb8fb2244daadaa59ebc2020b3c139e89", "source_url": "https://huggingface.co/sarvamai/sarvam-105b", "org": "sarvamai", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-03-06", "precision": "day", "url": "https://www.sarvam.ai/blogs/sarvam-30b-105b", "status": "verified_primary", "note": null }, "params": { "total_b": 105, "active_b": 10.3, "moe": true, "url": "https://huggingface.co/sarvamai/sarvam-105b/raw/abe264bdb8fb2244daadaa59ebc2020b3c139e89/README.md", "active_url": "https://huggingface.co/sarvamai/sarvam-105b/raw/abe264bdb8fb2244daadaa59ebc2020b3c139e89/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 131072, "url": "https://huggingface.co/sarvamai/sarvam-105b/raw/abe264bdb8fb2244daadaa59ebc2020b3c139e89/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/sarvamai/sarvam-105b/blob/abe264bdb8fb2244daadaa59ebc2020b3c139e89/LICENSE" }, "quality": { "rank": 68, "aa": { "value": 8.7905508357489, "variant": "Sarvam 105B (High)", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/sarvam-105b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/sarvamai/sarvam-105b/raw/abe264bdb8fb2244daadaa59ebc2020b3c139e89/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 64212541632, "repo": "limegreenpeper1/sarvam-105B-GGUF", "revision": "00ce70823d6fd94644a7778a0e4ec5bc6cac067e", "url": "https://huggingface.co/limegreenpeper1/sarvam-105B-GGUF/tree/00ce70823d6fd94644a7778a0e4ec5bc6cac067e", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 14, "status": "measured_file_manifest" }, "GGUF_Q8_0": { "bytes": 112724452832, "repo": "limegreenpeper1/sarvam-105B-GGUF", "revision": "00ce70823d6fd94644a7778a0e4ec5bc6cac067e", "url": "https://huggingface.co/limegreenpeper1/sarvam-105B-GGUF/tree/00ce70823d6fd94644a7778a0e4ec5bc6cac067e", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 24, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_Q5_K_M", "GGUF_FP16", "GGUF_BF16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "nanbeige4.1-3b", "name": "Nanbeige4.1-3B", "hf_repo": "Nanbeige/Nanbeige4.1-3B", "hf_revision": "9c4555c37921c0982af8fffeee1e0a6e6b953c4f", "source_url": "https://huggingface.co/Nanbeige/Nanbeige4.1-3B", "org": "Nanbeige", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-02", "precision": "month", "url": "https://artificialanalysis.ai/models/nanbeige4-1-3b", "status": "verified_secondary_aa_release_metadata", "note": "AA explicitly reports release month; exact primary-source public release day remains unknown." }, "params": { "total_b": 3, "active_b": 3, "moe": false, "url": "https://huggingface.co/Nanbeige/Nanbeige4.1-3B/blob/9c4555c37921c0982af8fffeee1e0a6e6b953c4f/README.md", "active_url": null, "scope": "Nominal model class; HF stored tensor total separately records embedding-inclusive count." }, "arch_label": "Dense", "context": { "native": 262144, "url": "https://huggingface.co/Nanbeige/Nanbeige4.1-3B/blob/9c4555c37921c0982af8fffeee1e0a6e6b953c4f/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/Nanbeige/Nanbeige4.1-3B/blob/9c4555c37921c0982af8fffeee1e0a6e6b953c4f/LICENSE" }, "quality": { "rank": 69, "aa": { "value": 8.4012282586083, "variant": "Nanbeige4.1-3B", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/nanbeige4-1-3b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": 65536, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/Nanbeige/Nanbeige4.1-3B/blob/9c4555c37921c0982af8fffeee1e0a6e6b953c4f/config.json", "dtype": "FP16", "note": "Ordinary GQA cache estimate; actual allocation and padding vary by backend." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 2443113024, "repo": "mradermacher/Nanbeige4.1-3B-GGUF", "revision": "c2c9e50a6bffb92c194c286e90cdf794b5657b13", "url": "https://huggingface.co/mradermacher/Nanbeige4.1-3B-GGUF/tree/c2c9e50a6bffb92c194c286e90cdf794b5657b13", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 2825269824, "repo": "mradermacher/Nanbeige4.1-3B-GGUF", "revision": "c2c9e50a6bffb92c194c286e90cdf794b5657b13", "url": "https://huggingface.co/mradermacher/Nanbeige4.1-3B-GGUF/tree/c2c9e50a6bffb92c194c286e90cdf794b5657b13", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 4183948864, "repo": "mradermacher/Nanbeige4.1-3B-GGUF", "revision": "c2c9e50a6bffb92c194c286e90cdf794b5657b13", "url": "https://huggingface.co/mradermacher/Nanbeige4.1-3B-GGUF/tree/c2c9e50a6bffb92c194c286e90cdf794b5657b13", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 7871577664, "repo": "mradermacher/Nanbeige4.1-3B-GGUF", "revision": "c2c9e50a6bffb92c194c286e90cdf794b5657b13", "url": "https://huggingface.co/mradermacher/Nanbeige4.1-3B-GGUF/tree/c2c9e50a6bffb92c194c286e90cdf794b5657b13", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 2212994356, "repo": "mlx-community/Nanbeige4.1-3B-4bit", "revision": "4d6c41d88f1416777bbda005aad92798e56e5ba9", "url": "https://huggingface.co/mlx-community/Nanbeige4.1-3B-4bit/tree/4d6c41d88f1416777bbda005aad92798e56e5ba9", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 4179730124, "repo": "mlx-community/Nanbeige4.1-3B-8bit", "revision": "699b0e755a2841c4c2762f846d4eaf93542ac97c", "url": "https://huggingface.co/mlx-community/Nanbeige4.1-3B-8bit/tree/699b0e755a2841c4c2762f846d4eaf93542ac97c", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-02-18", "title": "vLLM openai docker - tool config", "url": "https://huggingface.co/Nanbeige/Nanbeige4.1-3B/discussions/21" }, { "platform": "Hugging Face", "date": "2026-02-16", "title": "Training Data and inference scripts with tool calling , websearch and so on plus training scripts", "url": "https://huggingface.co/Nanbeige/Nanbeige4.1-3B/discussions/16" } ], "notes": [ "Exact public release day not verified from a primary announcement; repository creation and commit dates are deliberately not substituted." ] }, { "slug": "lfm2.5-2.6b", "name": "LFM2.5-2.6B", "hf_repo": "LiquidAI/LFM2.5-2.6B", "hf_revision": "654f9463ce32b05d0429d76fe1f580b27d4c1ac0", "source_url": "https://huggingface.co/LiquidAI/LFM2.5-2.6B", "org": "LiquidAI", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08-04", "precision": "day", "url": "https://www.liquid.ai/blog/lfm2-5-2-6b", "status": "verified_primary", "note": "Publisher explicitly says available today; resolves conflicting aggregator days." }, "params": { "total_b": 2.69, "active_b": 2.69, "moe": false, "url": "https://huggingface.co/LiquidAI/LFM2.5-2.6B/blob/654f9463ce32b05d0429d76fe1f580b27d4c1ac0/README.md", "active_url": null, "scope": "Publisher model-card count." }, "arch_label": "Dense · hybrid linear attn", "context": { "native": 131072, "url": "https://huggingface.co/LiquidAI/LFM2.5-2.6B/blob/654f9463ce32b05d0429d76fe1f580b27d4c1ac0/config.json", "extended": null }, "license": { "value": "lfm1.0", "url": "https://huggingface.co/LiquidAI/LFM2.5-2.6B/blob/654f9463ce32b05d0429d76fe1f580b27d4c1ac0/LICENSE" }, "quality": { "rank": 70, "aa": { "value": 8.38708534765242, "variant": "LFM2.5-2.6B", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/lfm2-5-2-6b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": 16384, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/LiquidAI/LFM2.5-2.6B/blob/654f9463ce32b05d0429d76fe1f580b27d4c1ac0/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 1674455040, "repo": "LiquidAI/LFM2.5-2.6B-GGUF", "revision": "e7caca5d835a3901a8e0d63e94009429bafafdfc", "url": "https://huggingface.co/LiquidAI/LFM2.5-2.6B-GGUF/tree/e7caca5d835a3901a8e0d63e94009429bafafdfc", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 1939744768, "repo": "LiquidAI/LFM2.5-2.6B-GGUF", "revision": "e7caca5d835a3901a8e0d63e94009429bafafdfc", "url": "https://huggingface.co/LiquidAI/LFM2.5-2.6B-GGUF/tree/e7caca5d835a3901a8e0d63e94009429bafafdfc", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 2874779648, "repo": "LiquidAI/LFM2.5-2.6B-GGUF", "revision": "e7caca5d835a3901a8e0d63e94009429bafafdfc", "url": "https://huggingface.co/LiquidAI/LFM2.5-2.6B-GGUF/tree/e7caca5d835a3901a8e0d63e94009429bafafdfc", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 5403158528, "repo": "LiquidAI/LFM2.5-2.6B-GGUF", "revision": "e7caca5d835a3901a8e0d63e94009429bafafdfc", "url": "https://huggingface.co/LiquidAI/LFM2.5-2.6B-GGUF/tree/e7caca5d835a3901a8e0d63e94009429bafafdfc", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 1583152892, "repo": "LiquidAI/LFM2.5-2.6B-MLX-4bit", "revision": "04efa23776ce61ec34ec95ec34c859854c89542b", "url": "https://huggingface.co/LiquidAI/LFM2.5-2.6B-MLX-4bit/tree/04efa23776ce61ec34ec95ec34c859854c89542b", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "mixed": true, "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 2866086056, "repo": "LiquidAI/LFM2.5-2.6B-MLX-8bit", "revision": "b372ebbb518c0e81617e25d8824427dd9ee1f08c", "url": "https://huggingface.co/LiquidAI/LFM2.5-2.6B-MLX-8bit/tree/b372ebbb518c0e81617e25d8824427dd9ee1f08c", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-08-27", "title": "<|tool_call_start|> and <|tool_call_end|> The biggest problem with this model.", "url": "https://huggingface.co/LiquidAI/LFM2.5-2.6B/discussions/18" }, { "platform": "Hugging Face", "date": "2026-09-06", "title": "Internal model structure preview", "url": "https://huggingface.co/LiquidAI/LFM2.5-2.6B/discussions/20" } ], "notes": [] }, { "slug": "falcon-h1r-7b", "name": "Falcon-H1R-7B", "hf_repo": "tiiuae/Falcon-H1R-7B", "hf_revision": "a6f74bf181389908efd6970878d7ee2b42f5d417", "source_url": "https://huggingface.co/tiiuae/Falcon-H1R-7B", "org": "tiiuae", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-01-05", "precision": "day", "url": "https://www.tii.ae/news/tii-launches-falcon-reasoning-best-7b-ai-model-globally-also-outperforms-larger-models", "status": "verified_primary", "note": "Official dated launch." }, "params": { "total_b": 7, "active_b": 7, "moe": false, "url": "https://huggingface.co/tiiuae/Falcon-H1R-7B/blob/a6f74bf181389908efd6970878d7ee2b42f5d417/README.md", "active_url": null, "scope": "Nominal class; hybrid Transformer/Mamba, not MoE." }, "arch_label": "Dense", "context": { "native": 262144, "url": "https://huggingface.co/tiiuae/Falcon-H1R-7B/blob/a6f74bf181389908efd6970878d7ee2b42f5d417/config.json", "extended": null }, "license": { "value": "falcon-llm-license", "url": "https://falconllm.tii.ae/falcon-terms-and-conditions.html" }, "quality": { "rank": 71, "aa": { "value": 7.82776928077145, "variant": "Falcon-H1R-7B", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/falcon-h1r-7b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": 45056, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/tiiuae/Falcon-H1R-7B/blob/a6f74bf181389908efd6970878d7ee2b42f5d417/config.json", "dtype": "FP16", "note": "Ordinary GQA cache estimate; actual allocation and padding vary by backend." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 4598343712, "repo": "tiiuae/Falcon-H1R-7B-GGUF", "revision": "2dc053e015a9e3c5b954aa81e00aaed24bef830f", "url": "https://huggingface.co/tiiuae/Falcon-H1R-7B-GGUF/tree/2dc053e015a9e3c5b954aa81e00aaed24bef830f", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 5390489632, "repo": "tiiuae/Falcon-H1R-7B-GGUF", "revision": "2dc053e015a9e3c5b954aa81e00aaed24bef830f", "url": "https://huggingface.co/tiiuae/Falcon-H1R-7B-GGUF/tree/2dc053e015a9e3c5b954aa81e00aaed24bef830f", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 8069003296, "repo": "tiiuae/Falcon-H1R-7B-GGUF", "revision": "2dc053e015a9e3c5b954aa81e00aaed24bef830f", "url": "https://huggingface.co/tiiuae/Falcon-H1R-7B-GGUF/tree/2dc053e015a9e3c5b954aa81e00aaed24bef830f", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 15179424128, "repo": "tiiuae/Falcon-H1R-7B-GGUF", "revision": "2dc053e015a9e3c5b954aa81e00aaed24bef830f", "url": "https://huggingface.co/tiiuae/Falcon-H1R-7B-GGUF/tree/2dc053e015a9e3c5b954aa81e00aaed24bef830f", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 4268829059, "repo": "mlx-community/Falcon-H1R-7B-4bit", "revision": "0e3ed6b0e4de5581f22500d2cfdc58e6f37568c6", "url": "https://huggingface.co/mlx-community/Falcon-H1R-7B-4bit/tree/0e3ed6b0e4de5581f22500d2cfdc58e6f37568c6", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 8061053079, "repo": "mlx-community/Falcon-H1R-7B-8bit", "revision": "2c25ee05aa8f3bbc78aad225a86d4d906da7b65e", "url": "https://huggingface.co/mlx-community/Falcon-H1R-7B-8bit/tree/2c25ee05aa8f3bbc78aad225a86d4d906da7b65e", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 2, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-01-19", "title": "Function call and tool usage with Open WebUI", "url": "https://huggingface.co/tiiuae/Falcon-H1R-7B/discussions/17" }, { "platform": "Hugging Face", "date": "2026-09-13", "title": "How to control reasoning effort in Falcon H1R 7B?", "url": "https://huggingface.co/tiiuae/Falcon-H1R-7B/discussions/20" } ], "notes": [] }, { "slug": "gemma-4-e2b", "name": "Gemma 4 E2B", "hf_repo": "google/gemma-4-E2B-it", "hf_revision": "3e22461f65e89153144f8adb70e3b8c2cc9845a7", "source_url": "https://huggingface.co/google/gemma-4-E2B-it", "org": "google", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-04-02", "precision": "day", "url": "https://blog.google/innovation-and-ai/technology/developers-tools/gemma-4/", "status": "verified_primary", "note": "Public launch announcement; distinguish earlier repository creation." }, "params": { "total_b": 5.1, "active_b": 5.1, "moe": false, "url": "https://huggingface.co/google/gemma-4-E2B-it/blob/3e22461f65e89153144f8adb70e3b8c2cc9845a7/README.md", "active_url": null, "scope": "2.3B effective; PLE embeddings make total 5.1B. Not a 2B stored-weight model." }, "arch_label": "Dense · sliding-window attn", "context": { "native": 131072, "url": "https://huggingface.co/google/gemma-4-E2B-it/blob/3e22461f65e89153144f8adb70e3b8c2cc9845a7/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://ai.google.dev/gemma/docs/gemma_4_license" }, "quality": { "rank": 72, "aa": { "value": 7.77445525187484, "variant": "Gemma 4 E2B (Reasoning)", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/gemma-4-e2b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/google/gemma-4-E2B-it/blob/3e22461f65e89153144f8adb70e3b8c2cc9845a7/config.json", "dtype": "FP16", "note": "Architecture needs backend-specific treatment: shared K/V or cross-layer cache, looped layers, MoVA, MLA/KDA or diffusion. No naive all-layers GQA estimate is supplied." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 3106738272, "repo": "unsloth/gemma-4-E2B-it-GGUF", "revision": "0314792d7f1f7e229411f620751375812bb9faf2", "url": "https://huggingface.co/unsloth/gemma-4-E2B-it-GGUF/tree/0314792d7f1f7e229411f620751375812bb9faf2", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 986833728, "url": "https://huggingface.co/unsloth/gemma-4-E2B-it-GGUF/blob/0314792d7f1f7e229411f620751375812bb9faf2/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 985654080, "url": "https://huggingface.co/unsloth/gemma-4-E2B-it-GGUF/blob/0314792d7f1f7e229411f620751375812bb9faf2/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1903027008, "url": "https://huggingface.co/unsloth/gemma-4-E2B-it-GGUF/blob/0314792d7f1f7e229411f620751375812bb9faf2/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 3356037216, "repo": "unsloth/gemma-4-E2B-it-GGUF", "revision": "0314792d7f1f7e229411f620751375812bb9faf2", "url": "https://huggingface.co/unsloth/gemma-4-E2B-it-GGUF/tree/0314792d7f1f7e229411f620751375812bb9faf2", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 986833728, "url": "https://huggingface.co/unsloth/gemma-4-E2B-it-GGUF/blob/0314792d7f1f7e229411f620751375812bb9faf2/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 985654080, "url": "https://huggingface.co/unsloth/gemma-4-E2B-it-GGUF/blob/0314792d7f1f7e229411f620751375812bb9faf2/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1903027008, "url": "https://huggingface.co/unsloth/gemma-4-E2B-it-GGUF/blob/0314792d7f1f7e229411f620751375812bb9faf2/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 5048352864, "repo": "unsloth/gemma-4-E2B-it-GGUF", "revision": "0314792d7f1f7e229411f620751375812bb9faf2", "url": "https://huggingface.co/unsloth/gemma-4-E2B-it-GGUF/tree/0314792d7f1f7e229411f620751375812bb9faf2", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 986833728, "url": "https://huggingface.co/unsloth/gemma-4-E2B-it-GGUF/blob/0314792d7f1f7e229411f620751375812bb9faf2/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 985654080, "url": "https://huggingface.co/unsloth/gemma-4-E2B-it-GGUF/blob/0314792d7f1f7e229411f620751375812bb9faf2/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1903027008, "url": "https://huggingface.co/unsloth/gemma-4-E2B-it-GGUF/blob/0314792d7f1f7e229411f620751375812bb9faf2/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 3550670554, "repo": "mlx-community/gemma-4-E2B-it-4bit", "revision": "238767527555cb75a05732a84dff5d6ba0dd6809", "url": "https://huggingface.co/mlx-community/gemma-4-E2B-it-4bit/tree/238767527555cb75a05732a84dff5d6ba0dd6809", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 5866581158, "repo": "mlx-community/gemma-4-E2B-it-8bit", "revision": "03dcf209f3f549b4075e7191e77cf69b3d48e1b2", "url": "https://huggingface.co/mlx-community/gemma-4-E2B-it-8bit/tree/03dcf209f3f549b4075e7191e77cf69b3d48e1b2", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 2, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-09-21", "title": "[Exploration] Applying Native Dual-Regime & KV-Cache Handoff to this model variant?", "url": "https://huggingface.co/google/gemma-4-E2B-it/discussions/48" }, { "platform": "Hugging Face", "date": "2026-06-10", "title": "Chat template may re-inject prior-turn reasoning during multi-turn tool use → repetition loops", "url": "https://huggingface.co/google/gemma-4-E2B-it/discussions/36" } ], "notes": [] }, { "slug": "step3-vl-10b", "name": "Step3 VL 10B", "hf_repo": "stepfun-ai/Step3-VL-10B", "hf_revision": "5026053b0c2f5dfaa08fc2d149384162c3c8bca1", "source_url": "https://huggingface.co/stepfun-ai/Step3-VL-10B", "org": "stepfun-ai", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-01", "precision": "month", "url": "https://artificialanalysis.ai/models/step-3-vl-10b", "status": "verified_secondary_aa_release_metadata", "note": "AA explicitly reports release month; exact primary-source public release day remains unknown." }, "params": { "total_b": 10, "active_b": 10, "moe": false, "url": "https://huggingface.co/stepfun-ai/Step3-VL-10B/blob/5026053b0c2f5dfaa08fc2d149384162c3c8bca1/README.md", "active_url": null, "scope": "Publisher total includes 1.8B visual encoder." }, "arch_label": "Dense", "context": { "native": 65536, "url": "https://huggingface.co/stepfun-ai/Step3-VL-10B/blob/5026053b0c2f5dfaa08fc2d149384162c3c8bca1/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/stepfun-ai/Step3-VL-10B/blob/5026053b0c2f5dfaa08fc2d149384162c3c8bca1/LICENSE" }, "quality": { "rank": 73, "aa": { "value": 7.69391371030502, "variant": "Step3 VL 10B", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/step-3-vl-10b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": 147456, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/stepfun-ai/Step3-VL-10B/blob/5026053b0c2f5dfaa08fc2d149384162c3c8bca1/config.json", "dtype": "FP16", "note": "Ordinary GQA cache estimate; actual allocation and padding vary by backend." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 5027784512, "repo": "JamePeng2023/Step3-VL-10B-GGUF", "revision": "f0bc308167bb03e463e10ba8dfaa4879092d0e84", "url": "https://huggingface.co/JamePeng2023/Step3-VL-10B-GGUF/tree/f0bc308167bb03e463e10ba8dfaa4879092d0e84", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-Step3-VL-10b-F16.gguf", "bytes": 3972829344, "url": "https://huggingface.co/JamePeng2023/Step3-VL-10B-GGUF/blob/f0bc308167bb03e463e10ba8dfaa4879092d0e84/mmproj-Step3-VL-10b-F16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 5851113280, "repo": "Vastined/Step3-VL-10B-GGUF", "revision": "8910c4151e1027034f587024fdd145d542cacfee", "url": "https://huggingface.co/Vastined/Step3-VL-10B-GGUF/tree/8910c4151e1027034f587024fdd145d542cacfee", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-Step3-VL-10B-BF16.gguf", "bytes": 4399308960, "url": "https://huggingface.co/Vastined/Step3-VL-10B-GGUF/blob/8910c4151e1027034f587024fdd145d542cacfee/mmproj-Step3-VL-10B-BF16.gguf" }, { "path": "mmproj-Step3-VL-10B-F16.gguf", "bytes": 3972829344, "url": "https://huggingface.co/Vastined/Step3-VL-10B-GGUF/blob/8910c4151e1027034f587024fdd145d542cacfee/mmproj-Step3-VL-10B-F16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 8709519168, "repo": "JamePeng2023/Step3-VL-10B-GGUF", "revision": "f0bc308167bb03e463e10ba8dfaa4879092d0e84", "url": "https://huggingface.co/JamePeng2023/Step3-VL-10B-GGUF/tree/f0bc308167bb03e463e10ba8dfaa4879092d0e84", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-Step3-VL-10b-F16.gguf", "bytes": 3972829344, "url": "https://huggingface.co/JamePeng2023/Step3-VL-10B-GGUF/blob/f0bc308167bb03e463e10ba8dfaa4879092d0e84/mmproj-Step3-VL-10b-F16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 16388044608, "repo": "seanbailey518/Step3-VL-10B-GGUF", "revision": "4e88ea55358ca8091e495a59414fd4a5d43d6932", "url": "https://huggingface.co/seanbailey518/Step3-VL-10B-GGUF/tree/4e88ea55358ca8091e495a59414fd4a5d43d6932", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-Step3-VL-10b-F16.gguf", "bytes": 3964522624, "url": "https://huggingface.co/seanbailey518/Step3-VL-10B-GGUF/blob/4e88ea55358ca8091e495a59414fd4a5d43d6932/mmproj-Step3-VL-10b-F16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "MLX_4bit", "MLX_8bit" ], "community": [ { "platform": "Hugging Face", "date": "2026-01-24", "title": "tool calls sometimes fail on vllm", "url": "https://huggingface.co/stepfun-ai/Step3-VL-10B/discussions/19" }, { "platform": "Hugging Face", "date": "2026-01-22", "title": "Not support flash_attn_2？", "url": "https://huggingface.co/stepfun-ai/Step3-VL-10B/discussions/17" } ], "notes": [ "Exact public release day not verified from a primary announcement; repository creation and commit dates are deliberately not substituted." ] }, { "slug": "lfm2.5-8b-a1b", "name": "LFM2.5-8B-A1B", "hf_repo": "LiquidAI/LFM2.5-8B-A1B", "hf_revision": "5dd22602c2e9f6a097b1de4c4efe0658b605015c", "source_url": "https://huggingface.co/LiquidAI/LFM2.5-8B-A1B", "org": "LiquidAI", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-05-28", "precision": "day", "url": "https://www.liquid.ai/blog/lfm2-5-8b-a1b", "status": "verified_primary", "note": "Publisher announcement." }, "params": { "total_b": 8.3, "active_b": 1.5, "moe": true, "url": "https://huggingface.co/LiquidAI/LFM2.5-8B-A1B/blob/5dd22602c2e9f6a097b1de4c4efe0658b605015c/README.md", "active_url": "https://huggingface.co/LiquidAI/LFM2.5-8B-A1B/blob/5dd22602c2e9f6a097b1de4c4efe0658b605015c/README.md", "scope": "Live card says 8.3B total / 1.5B active despite A1B name." }, "arch_label": "MoE · hybrid linear attn", "context": { "native": 128000, "url": "https://huggingface.co/LiquidAI/LFM2.5-8B-A1B/blob/5dd22602c2e9f6a097b1de4c4efe0658b605015c/config.json", "extended": null }, "license": { "value": "lfm1.0", "url": "https://huggingface.co/LiquidAI/LFM2.5-8B-A1B/blob/5dd22602c2e9f6a097b1de4c4efe0658b605015c/LICENSE" }, "quality": { "rank": 74, "aa": { "value": 7.21643567343487, "variant": "LFM2.5-8B-A1B", "estimated": true, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/lfm2-5-8b-a1b", "observed_at": "2026-10-17", "status": "publisher_estimated_index" }, "arena": null }, "kv": { "grow": 12288, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/LiquidAI/LFM2.5-8B-A1B/blob/5dd22602c2e9f6a097b1de4c4efe0658b605015c/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 5155564768, "repo": "LiquidAI/LFM2.5-8B-A1B-GGUF", "revision": "49c14831707011e64d70b2ebd8462ba08d608434", "url": "https://huggingface.co/LiquidAI/LFM2.5-8B-A1B-GGUF/tree/49c14831707011e64d70b2ebd8462ba08d608434", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 6030339296, "repo": "LiquidAI/LFM2.5-8B-A1B-GGUF", "revision": "49c14831707011e64d70b2ebd8462ba08d608434", "url": "https://huggingface.co/LiquidAI/LFM2.5-8B-A1B-GGUF/tree/49c14831707011e64d70b2ebd8462ba08d608434", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 9010195680, "repo": "LiquidAI/LFM2.5-8B-A1B-GGUF", "revision": "49c14831707011e64d70b2ebd8462ba08d608434", "url": "https://huggingface.co/LiquidAI/LFM2.5-8B-A1B-GGUF/tree/49c14831707011e64d70b2ebd8462ba08d608434", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 16947260640, "repo": "LiquidAI/LFM2.5-8B-A1B-GGUF", "revision": "49c14831707011e64d70b2ebd8462ba08d608434", "url": "https://huggingface.co/LiquidAI/LFM2.5-8B-A1B-GGUF/tree/49c14831707011e64d70b2ebd8462ba08d608434", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 4834029941, "repo": "LiquidAI/LFM2.5-8B-A1B-MLX-4bit", "revision": "2e92b640a63d47ad4dcf81a19a366b902356b3bc", "url": "https://huggingface.co/LiquidAI/LFM2.5-8B-A1B-MLX-4bit/tree/2e92b640a63d47ad4dcf81a19a366b902356b3bc", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "mixed": true, "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 9001595267, "repo": "LiquidAI/LFM2.5-8B-A1B-MLX-8bit", "revision": "c663adcc7d9d82abc4d482cb4f226e0a7eed3170", "url": "https://huggingface.co/LiquidAI/LFM2.5-8B-A1B-MLX-8bit/tree/c663adcc7d9d82abc4d482cb4f226e0a7eed3170", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 2, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-09-20", "title": "Tool call sometimes emitted as plain JSON in `content` (missing `<|tool_call_start|>` after ``) — llama.cpp + pi agent", "url": "https://huggingface.co/LiquidAI/LFM2.5-8B-A1B/discussions/36" }, { "platform": "Hugging Face", "date": "2026-07-02", "title": "Italian Support", "url": "https://huggingface.co/LiquidAI/LFM2.5-8B-A1B/discussions/27" } ], "notes": [] }, { "slug": "nemotron-3-nano-4b", "name": "NVIDIA Nemotron 3 Nano 4B", "hf_repo": "nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16", "hf_revision": "dfaf35de3e30f1867dd8dbc38a7fc9fb52d3914f", "source_url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16", "org": "nvidia", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-03-16", "precision": "day", "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16/blob/dfaf35de3e30f1867dd8dbc38a7fc9fb52d3914f/README.md", "status": "verified_primary", "note": "Model card explicit Release Date." }, "params": { "total_b": 3.97, "active_b": 3.97, "moe": false, "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16/blob/dfaf35de3e30f1867dd8dbc38a7fc9fb52d3914f/README.md", "active_url": null, "scope": "Publisher card 3.97 x 10^9." }, "arch_label": "Dense · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16/blob/dfaf35de3e30f1867dd8dbc38a7fc9fb52d3914f/config.json", "extended": null }, "license": { "value": "nvidia-nemotron-open-model-license", "url": "https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-nemotron-open-model-license/" }, "quality": { "rank": 75, "aa": { "value": 4.76910061591993, "variant": "NVIDIA Nemotron 3 Nano 4B", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/nvidia-nemotron-3-nano-4b", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": null }, "kv": { "grow": 16384, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16/blob/dfaf35de3e30f1867dd8dbc38a7fc9fb52d3914f/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 2837072864, "repo": "nvidia/NVIDIA-Nemotron-3-Nano-4B-GGUF", "revision": "ba223d14e45525f7fae81db77ea8cabeb2fc6c25", "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-4B-GGUF/tree/ba223d14e45525f7fae81db77ea8cabeb2fc6c25", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 3159586464, "repo": "unsloth/NVIDIA-Nemotron-3-Nano-4B-GGUF", "revision": "8e81be55c5aa3d63bb82b6ceec62d50805d9e1bb", "url": "https://huggingface.co/unsloth/NVIDIA-Nemotron-3-Nano-4B-GGUF/tree/8e81be55c5aa3d63bb82b6ceec62d50805d9e1bb", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 4233679008, "repo": "unsloth/NVIDIA-Nemotron-3-Nano-4B-GGUF", "revision": "8e81be55c5aa3d63bb82b6ceec62d50805d9e1bb", "url": "https://huggingface.co/unsloth/NVIDIA-Nemotron-3-Nano-4B-GGUF/tree/8e81be55c5aa3d63bb82b6ceec62d50805d9e1bb", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 2237078664, "repo": "mlx-community/NVIDIA-Nemotron-3-Nano-4B-4bit", "revision": "c4d79ba1901d99806ef757642a552acebb851a35", "url": "https://huggingface.co/mlx-community/NVIDIA-Nemotron-3-Nano-4B-4bit/tree/c4d79ba1901d99806ef757642a552acebb851a35", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 4223195521, "repo": "mlx-community/NVIDIA-Nemotron-3-Nano-4B-8bit", "revision": "0f1404e03ad62eb769874559e8066c33db6455ce", "url": "https://huggingface.co/mlx-community/NVIDIA-Nemotron-3-Nano-4B-8bit/tree/0f1404e03ad62eb769874559e8066c33db6455ce", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-03-29", "title": "Does the AIME benchmark report use a tool or not?", "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16/discussions/3" }, { "platform": "Hugging Face", "date": "2026-06-07", "title": "TemporalMesh Transformer: 29.4 PPL at 48% compute — beats Mamba, new open-source architecture", "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16/discussions/7" } ], "notes": [] }, { "slug": "kolibri-1", "name": "Aleph Alpha Kolibri-1", "hf_repo": "Aleph-Alpha/Kolibri-1", "hf_revision": "452723312f300bf9542cd425f9d39bf832c2af40", "source_url": "https://huggingface.co/Aleph-Alpha/Kolibri-1", "org": "Aleph-Alpha", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-10-03", "precision": "day", "url": "https://huggingface.co/Aleph-Alpha/Kolibri-1/raw/main/README.md", "status": "verified_primary", "note": null }, "params": { "total_b": 78, "active_b": 3.46, "moe": true, "url": "https://huggingface.co/Aleph-Alpha/Kolibri-1/raw/452723312f300bf9542cd425f9d39bf832c2af40/README.md", "active_url": "https://huggingface.co/Aleph-Alpha/Kolibri-1/raw/452723312f300bf9542cd425f9d39bf832c2af40/README.md", "scope": null }, "arch_label": "MoE · sliding-window attn", "context": { "native": 262144, "url": "https://huggingface.co/Aleph-Alpha/Kolibri-1/raw/452723312f300bf9542cd425f9d39bf832c2af40/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/Aleph-Alpha/Kolibri-1/blob/452723312f300bf9542cd425f9d39bf832c2af40/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 20480, "win_per_token": null, "win_tokens": null, "win_max": 42024960, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/Aleph-Alpha/Kolibri-1/raw/452723312f300bf9542cd425f9d39bf832c2af40/config.json", "dtype": "BF16", "note": "BF16/FP16 KV estimate for full-attention layers only, batch=1. Sliding-window terms and fixed recurrent state are separate; excludes alignment, graphs, buffers, vision and MTP/drafters." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 47454113472, "repo": "Hob-forge/Kolibri-1-GGUF", "revision": "40eaa5cf6cfca3afaf8b8832bac962652edb0bde", "url": "https://huggingface.co/Hob-forge/Kolibri-1-GGUF/tree/40eaa5cf6cfca3afaf8b8832bac962652edb0bde", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 1, "status": "measured_file_manifest" }, "GGUF_Q5_K_M": { "bytes": 55588769664, "repo": "Hob-forge/Kolibri-1-GGUF", "revision": "40eaa5cf6cfca3afaf8b8832bac962652edb0bde", "url": "https://huggingface.co/Hob-forge/Kolibri-1-GGUF/tree/40eaa5cf6cfca3afaf8b8832bac962652edb0bde", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 2, "status": "measured_file_manifest" }, "GGUF_Q8_0": { "bytes": 83135393696, "repo": "Hob-forge/Kolibri-1-GGUF", "revision": "40eaa5cf6cfca3afaf8b8832bac962652edb0bde", "url": "https://huggingface.co/Hob-forge/Kolibri-1-GGUF/tree/40eaa5cf6cfca3afaf8b8832bac962652edb0bde", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 2, "status": "measured_file_manifest" }, "MLX_4bit": { "bytes": 44332328498, "repo": "velaia/Kolibri-1-MLX-4bit", "revision": "3f5adf3fc8149f57738cc5a99f02ae26b601e7b0", "url": "https://huggingface.co/velaia/Kolibri-1-MLX-4bit/tree/3f5adf3fc8149f57738cc5a99f02ae26b601e7b0", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 9, "status": "measured_file_manifest", "mixed": true } }, "artifacts_missing": [ "GGUF_FP16", "GGUF_BF16", "MLX_8bit" ], "community": [], "notes": [ "Official source is native FP8, not FP16/BF16 just because config dtype says bfloat16." ] }, { "slug": "deepseek-v4-flash-0731", "name": "DeepSeek V4 Flash 0731", "hf_repo": "deepseek-ai/DeepSeek-V4-Flash-0731", "hf_revision": "7872f01b1d1fe23eabc4c98b48bffcef5a386062", "source_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731", "org": "deepseek-ai", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-07-31", "precision": "day", "url": "https://api-docs.deepseek.com/updates/", "status": "verified_primary", "note": null }, "params": { "total_b": null, "active_b": null, "moe": true, "url": null, "active_url": null, "scope": null }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731/raw/7872f01b1d1fe23eabc4c98b48bffcef5a386062/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731/blob/7872f01b1d1fe23eabc4c98b48bffcef5a386062/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731/raw/7872f01b1d1fe23eabc4c98b48bffcef5a386062/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "MLX_4bit": { "bytes": 154219072348, "repo": "ToPo-ToPo/DeepSeek-V4-Flash-0731-mlx-4bit", "revision": "c8b8f3fe9112cc5aa3b34d2aec9f771b68622902", "url": "https://huggingface.co/ToPo-ToPo/DeepSeek-V4-Flash-0731-mlx-4bit/tree/c8b8f3fe9112cc5aa3b34d2aec9f771b68622902", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 33, "status": "measured_file_manifest", "mixed": true } }, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_Q8_0", "GGUF_FP16", "GGUF_BF16", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "glm-4.7-flash", "name": "GLM-4.7-Flash", "hf_repo": "zai-org/GLM-4.7-Flash", "hf_revision": "7dd20894a642a0aa287e9827cb1a1f7f91386b67", "source_url": "https://huggingface.co/zai-org/GLM-4.7-Flash", "org": "zai-org", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-01", "precision": "month", "url": "https://artificialanalysis.ai/models/glm-4-7-flash", "status": "verified_secondary_aa_release_metadata", "note": "AA explicitly reports release month; exact primary-source public release day remains unknown." }, "params": { "total_b": 30, "active_b": 3, "moe": true, "url": "https://huggingface.co/zai-org/GLM-4.7-Flash/blob/7dd20894a642a0aa287e9827cb1a1f7f91386b67/README.md", "active_url": "https://huggingface.co/zai-org/GLM-4.7-Flash/blob/7dd20894a642a0aa287e9827cb1a1f7f91386b67/README.md", "scope": "Publisher nominal total/active count." }, "arch_label": "MoE", "context": { "native": 202752, "url": "https://huggingface.co/zai-org/GLM-4.7-Flash/blob/7dd20894a642a0aa287e9827cb1a1f7f91386b67/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/zai-org/GLM-4.7-Flash/blob/7dd20894a642a0aa287e9827cb1a1f7f91386b67/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/zai-org/GLM-4.7-Flash/blob/7dd20894a642a0aa287e9827cb1a1f7f91386b67/config.json", "dtype": "FP16", "note": "Compressed versus expanded MLA is backend dependent; candidate is not asserted as actual allocation." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 18312339808, "repo": "unsloth/GLM-4.7-Flash-GGUF", "revision": "0d32489ecb9db6d2a4fc93bd27ef01519f95474d", "url": "https://huggingface.co/unsloth/GLM-4.7-Flash-GGUF/tree/0d32489ecb9db6d2a4fc93bd27ef01519f95474d", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 21408850272, "repo": "unsloth/GLM-4.7-Flash-GGUF", "revision": "0d32489ecb9db6d2a4fc93bd27ef01519f95474d", "url": "https://huggingface.co/unsloth/GLM-4.7-Flash-GGUF/tree/0d32489ecb9db6d2a4fc93bd27ef01519f95474d", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 31842799968, "repo": "unsloth/GLM-4.7-Flash-GGUF", "revision": "0d32489ecb9db6d2a4fc93bd27ef01519f95474d", "url": "https://huggingface.co/unsloth/GLM-4.7-Flash-GGUF/tree/0d32489ecb9db6d2a4fc93bd27ef01519f95474d", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 16852441819, "repo": "mlx-community/GLM-4.7-Flash-4bit", "revision": "1454cffb1a21737e162f508e5bc70be9def89276", "url": "https://huggingface.co/mlx-community/GLM-4.7-Flash-4bit/tree/1454cffb1a21737e162f508e5bc70be9def89276", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 4, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 31820995674, "repo": "mlx-community/GLM-4.7-Flash-8bit", "revision": "b3a202c6df57f7297fb351486938952352dcd25a", "url": "https://huggingface.co/mlx-community/GLM-4.7-Flash-8bit/tree/b3a202c6df57f7297fb351486938952352dcd25a", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 7, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-02-16", "title": "Quantized GLM-4.7-Flash with llama.cpp and opencode.", "url": "https://huggingface.co/zai-org/GLM-4.7-Flash/discussions/66" }, { "platform": "Hugging Face", "date": "2026-02-12", "title": "Is FIM supported for code completion？", "url": "https://huggingface.co/zai-org/GLM-4.7-Flash/discussions/65" } ], "notes": [ "Exact public release day not verified from a primary announcement; repository creation and commit dates are deliberately not substituted." ] }, { "slug": "glm-5.2", "name": "GLM-5.2", "hf_repo": "zai-org/GLM-5.2", "hf_revision": "cf457fa734ab149ffef225f80893eb38c6ff5cdc", "source_url": "https://huggingface.co/zai-org/GLM-5.2", "org": "zai-org", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-06-16", "precision": "day", "url": "https://docs.z.ai/release-notes/new-released", "status": "verified_primary", "note": null }, "params": { "total_b": null, "active_b": null, "moe": true, "url": null, "active_url": null, "scope": null }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/zai-org/GLM-5.2/raw/cf457fa734ab149ffef225f80893eb38c6ff5cdc/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/zai-org/GLM-5.2/blob/cf457fa734ab149ffef225f80893eb38c6ff5cdc/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/zai-org/GLM-5.2/raw/cf457fa734ab149ffef225f80893eb38c6ff5cdc/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "GGUF_Q8_0": { "bytes": 801357672256, "repo": "unsloth/GLM-5.2-GGUF", "revision": "abc55e72527792c6e77069c99b4cb7de16fa9f23", "url": "https://huggingface.co/unsloth/GLM-5.2-GGUF/tree/abc55e72527792c6e77069c99b4cb7de16fa9f23", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 17, "status": "measured_file_manifest" }, "GGUF_BF16": { "bytes": 1507988023008, "repo": "unsloth/GLM-5.2-GGUF", "revision": "abc55e72527792c6e77069c99b4cb7de16fa9f23", "url": "https://huggingface.co/unsloth/GLM-5.2-GGUF/tree/abc55e72527792c6e77069c99b4cb7de16fa9f23", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 33, "status": "measured_file_manifest" }, "MLX_4bit": { "bytes": 418321321530, "repo": "mlx-community/GLM-5.2-4bit", "revision": "6832ff4c65e23ed5737d244de196d5a6024fe55b", "url": "https://huggingface.co/mlx-community/GLM-5.2-4bit/tree/6832ff4c65e23ed5737d244de196d5a6024fe55b", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 91, "status": "measured_file_manifest" }, "MLX_8bit": { "bytes": 789950255209, "repo": "pipenetwork/GLM-5.2-MLX-8bit", "revision": "26e36618698266d212394c943515e7d496baf9fb", "url": "https://huggingface.co/pipenetwork/GLM-5.2-MLX-8bit/tree/26e36618698266d212394c943515e7d496baf9fb", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 226, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_FP16" ], "community": [], "notes": [] }, { "slug": "hy4-preview", "name": "Hy4-preview", "hf_repo": "tencent/Hy4-preview", "hf_revision": "705d81ee51566a186d645b74c974d642ef2828fe", "source_url": "https://huggingface.co/tencent/Hy4-preview", "org": "tencent", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08-28", "precision": "day", "url": "https://www.tencent.com/tencent-releases-and-open-sources-tencent-hy4-preview/", "status": "verified_primary", "note": "Publisher explicitly released and open-sourced Hy4 preview." }, "params": { "total_b": 770, "active_b": 49, "moe": true, "url": "https://huggingface.co/tencent/Hy4-preview/raw/705d81ee51566a186d645b74c974d642ef2828fe/README.md", "active_url": "https://huggingface.co/tencent/Hy4-preview/raw/705d81ee51566a186d645b74c974d642ef2828fe/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/tencent/Hy4-preview/raw/705d81ee51566a186d645b74c974d642ef2828fe/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/tencent/Hy4-preview/blob/705d81ee51566a186d645b74c974d642ef2828fe/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/tencent/Hy4-preview/raw/705d81ee51566a186d645b74c974d642ef2828fe/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 467292398016, "repo": "AngelSlim/Hy4-preview-GGUF", "revision": "779242edccdedc2109a0b36b164263a88f015bfa", "url": "https://huggingface.co/AngelSlim/Hy4-preview-GGUF/tree/779242edccdedc2109a0b36b164263a88f015bfa", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 1, "status": "measured_file_manifest" }, "MLX_4bit": { "bytes": 433355075408, "repo": "mlx-community/Hy4-preview-4bit", "revision": "6c83660fe83ef0626bfbb70d7281cfb77d3f4245", "url": "https://huggingface.co/mlx-community/Hy4-preview-4bit/tree/6c83660fe83ef0626bfbb70d7281cfb77d3f4245", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 103, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_Q5_K_M", "GGUF_Q8_0", "GGUF_FP16", "GGUF_BF16", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "k2-horizon-0.9b", "name": "K2-Horizon-0.9B", "hf_repo": "IFM/K2-Horizon-0.9B", "hf_revision": "383cb66347be4d361138e8157c2f8779b7440752", "source_url": "https://huggingface.co/IFM/K2-Horizon-0.9B", "org": "IFM", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-09-03", "precision": "day", "url": "https://mbzuai.ac.ae/news-events/news/mbzuais-institute-foundation-models-launches-k2-horizon-the-worlds-largest-fully", "status": "verified_primary", "note": "Publisher launch announces six models available immediately, including this size." }, "params": { "total_b": null, "active_b": null, "moe": false, "url": null, "active_url": null, "scope": "0.9B is core model class; HF total is separate." }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/IFM/K2-Horizon-0.9B/blob/383cb66347be4d361138e8157c2f8779b7440752/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/IFM/K2-Horizon-0.9B/blob/383cb66347be4d361138e8157c2f8779b7440752/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 57344, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/IFM/K2-Horizon-0.9B/blob/383cb66347be4d361138e8157c2f8779b7440752/config.json", "dtype": "FP16", "note": "Ordinary GQA cache estimate; actual allocation and padding vary by backend." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 666184064, "repo": "IFM/K2-Horizon-0.9B-GGUF", "revision": "118965081cf1d876990219a1856a93a97b2668fa", "url": "https://huggingface.co/IFM/K2-Horizon-0.9B-GGUF/tree/118965081cf1d876990219a1856a93a97b2668fa", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 773482880, "repo": "IFM/K2-Horizon-0.9B-GGUF", "revision": "118965081cf1d876990219a1856a93a97b2668fa", "url": "https://huggingface.co/IFM/K2-Horizon-0.9B-GGUF/tree/118965081cf1d876990219a1856a93a97b2668fa", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 1148614016, "repo": "IFM/K2-Horizon-0.9B-GGUF", "revision": "118965081cf1d876990219a1856a93a97b2668fa", "url": "https://huggingface.co/IFM/K2-Horizon-0.9B-GGUF/tree/118965081cf1d876990219a1856a93a97b2668fa", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 2159435008, "repo": "mradermacher/K2-Horizon-0.9B-GGUF", "revision": "25cb9b673a2b1e7c80fde9db1930208629d7583f", "url": "https://huggingface.co/mradermacher/K2-Horizon-0.9B-GGUF/tree/25cb9b673a2b1e7c80fde9db1930208629d7583f", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 856819751, "repo": "mlx-community/K2-Horizon-0.9B-4bit", "revision": "4b5544bf46094c4b81e4bdd5a23d52b9b7e2fa75", "url": "https://huggingface.co/mlx-community/K2-Horizon-0.9B-4bit/tree/4b5544bf46094c4b81e4bdd5a23d52b9b7e2fa75", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "mixed": true, "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 1145833554, "repo": "mlx-community/K2-Horizon-0.9B-8bit", "revision": "72776797ed87450db16db4c345b57b54465daa03", "url": "https://huggingface.co/mlx-community/K2-Horizon-0.9B-8bit/tree/72776797ed87450db16db4c345b57b54465daa03", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-10-12", "title": "Hyperparameters Used During Training Are Not Documented", "url": "https://huggingface.co/IFM/K2-Horizon-0.9B/discussions/13" }, { "platform": "Hugging Face", "date": "2026-10-07", "title": "K2-Horizon public release artifacts: four template and documentation observations", "url": "https://huggingface.co/IFM/K2-Horizon-0.9B/discussions/12" } ], "notes": [] }, { "slug": "kimi-k2.6", "name": "Kimi K2.6", "hf_repo": "moonshotai/Kimi-K2.6", "hf_revision": "7eb5002f6aadc958aed6a9177b7ed26bb94011bb", "source_url": "https://huggingface.co/moonshotai/Kimi-K2.6", "org": "moonshotai", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-04-20", "precision": "day", "url": "https://www.kimi.ai/help/agent/agent-overview", "status": "verified_primary", "note": "Official retrospective release timeline explicitly says released and open-sourced." }, "params": { "total_b": 1000, "active_b": 32, "moe": true, "url": "https://huggingface.co/moonshotai/Kimi-K2.6/raw/7eb5002f6aadc958aed6a9177b7ed26bb94011bb/README.md", "active_url": "https://huggingface.co/moonshotai/Kimi-K2.6/raw/7eb5002f6aadc958aed6a9177b7ed26bb94011bb/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 262144, "url": "https://huggingface.co/moonshotai/Kimi-K2.6/raw/7eb5002f6aadc958aed6a9177b7ed26bb94011bb/config.json", "extended": null }, "license": { "value": "modified-mit", "url": "https://huggingface.co/moonshotai/Kimi-K2.6/blob/7eb5002f6aadc958aed6a9177b7ed26bb94011bb/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/moonshotai/Kimi-K2.6/raw/7eb5002f6aadc958aed6a9177b7ed26bb94011bb/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "GGUF_BF16": { "bytes": 2053155814880, "repo": "unsloth/Kimi-K2.6-GGUF", "revision": "47c7cab1e440dd9fcfc57c469e4737983408a6f2", "url": "https://huggingface.co/unsloth/Kimi-K2.6-GGUF/tree/47c7cab1e440dd9fcfc57c469e4737983408a6f2", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 46, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 953927232, "url": "https://huggingface.co/unsloth/Kimi-K2.6-GGUF/resolve/47c7cab1e440dd9fcfc57c469e4737983408a6f2/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 952572480, "url": "https://huggingface.co/unsloth/Kimi-K2.6-GGUF/resolve/47c7cab1e440dd9fcfc57c469e4737983408a6f2/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1884595776, "url": "https://huggingface.co/unsloth/Kimi-K2.6-GGUF/resolve/47c7cab1e440dd9fcfc57c469e4737983408a6f2/mmproj-F32.gguf" } ] }, "MLX_4bit": { "bytes": 657623433662, "repo": "aidiffuser/Kimi-K2.6-MLX-4bit", "revision": "f3526deda33ce0363fc929dad500638c2ae620ac", "url": "https://huggingface.co/aidiffuser/Kimi-K2.6-MLX-4bit/tree/f3526deda33ce0363fc929dad500638c2ae620ac", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 182, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_Q8_0", "GGUF_FP16", "MLX_8bit" ], "community": [], "notes": [] }, { "slug": "lfm2.5-24b-a2b", "name": "LFM2.5-24B-A2B", "hf_repo": "LiquidAI/LFM2.5-24B-A2B", "hf_revision": "1769516ac588c4c558089c94f27d96b0dc67b183", "source_url": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B", "org": "LiquidAI", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-10-13", "precision": "day", "url": "https://www.liquid.ai/blog/lfm2-5-24b-a2b", "status": "verified_primary", "note": "Publisher announcement." }, "params": { "total_b": 24, "active_b": 2, "moe": true, "url": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B/blob/1769516ac588c4c558089c94f27d96b0dc67b183/README.md", "active_url": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B/blob/1769516ac588c4c558089c94f27d96b0dc67b183/README.md", "scope": "Publisher nominal total/active counts." }, "arch_label": "MoE · hybrid linear attn", "context": { "native": 131072, "url": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B/blob/1769516ac588c4c558089c94f27d96b0dc67b183/config.json", "extended": null }, "license": { "value": "lfm1.0", "url": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B/blob/1769516ac588c4c558089c94f27d96b0dc67b183/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 20480, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B/blob/1769516ac588c4c558089c94f27d96b0dc67b183/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 14526258464, "repo": "LiquidAI/LFM2.5-24B-A2B-GGUF", "revision": "0912d33141f77e10705345a6732bfd83f0252fdc", "url": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B-GGUF/tree/0912d33141f77e10705345a6732bfd83f0252fdc", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 17029602592, "repo": "LiquidAI/LFM2.5-24B-A2B-GGUF", "revision": "0912d33141f77e10705345a6732bfd83f0252fdc", "url": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B-GGUF/tree/0912d33141f77e10705345a6732bfd83f0252fdc", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 25493732640, "repo": "LiquidAI/LFM2.5-24B-A2B-GGUF", "revision": "0912d33141f77e10705345a6732bfd83f0252fdc", "url": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B-GGUF/tree/0912d33141f77e10705345a6732bfd83f0252fdc", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 47962094880, "repo": "LiquidAI/LFM2.5-24B-A2B-GGUF", "revision": "0912d33141f77e10705345a6732bfd83f0252fdc", "url": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B-GGUF/tree/0912d33141f77e10705345a6732bfd83f0252fdc", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 13567294335, "repo": "LiquidAI/LFM2.5-24B-A2B-MLX-4bit", "revision": "667408d494bde191c2cb9dd4f4549827816b9275", "url": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B-MLX-4bit/tree/667408d494bde191c2cb9dd4f4549827816b9275", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 3, "status": "verified_hf_file_sizes", "mixed": true, "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 25484884997, "repo": "LiquidAI/LFM2.5-24B-A2B-MLX-8bit", "revision": "67ab417dd02d0c9545802a726df668620eae049d", "url": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B-MLX-8bit/tree/67ab417dd02d0c9545802a726df668620eae049d", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 5, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-10-14", "title": "LFM2.5-24B-A2B, vram requirements", "url": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B/discussions/6" }, { "platform": "Hugging Face", "date": "2026-10-13", "title": "Blog post error", "url": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B/discussions/3" } ], "notes": [] }, { "slug": "ling-3-0-flash", "name": "Ling-3.0-flash", "hf_repo": "inclusionAI/Ling-3.0-flash", "hf_revision": "fcc60a52b34c69c6271d0ac8a8cb0815020b6a43", "source_url": "https://huggingface.co/inclusionAI/Ling-3.0-flash", "org": "inclusionAI", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08-07", "precision": "day", "url": "https://www.ant-ling.com/en", "status": "verified_primary", "note": "Publisher explicitly labels open-source release; earlier July24 was API announcement." }, "params": { "total_b": 124, "active_b": 5.1, "moe": true, "url": "https://huggingface.co/inclusionAI/Ling-3.0-flash/raw/fcc60a52b34c69c6271d0ac8a8cb0815020b6a43/README.md", "active_url": "https://huggingface.co/inclusionAI/Ling-3.0-flash/raw/fcc60a52b34c69c6271d0ac8a8cb0815020b6a43/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 262144, "url": "https://huggingface.co/inclusionAI/Ling-3.0-flash/raw/fcc60a52b34c69c6271d0ac8a8cb0815020b6a43/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/inclusionAI/Ling-3.0-flash/blob/fcc60a52b34c69c6271d0ac8a8cb0815020b6a43/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/inclusionAI/Ling-3.0-flash/raw/fcc60a52b34c69c6271d0ac8a8cb0815020b6a43/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 77804990144, "repo": "bartowski/Ling-3.0-flash-GGUF", "revision": "7e1a1243566501d52af98e9aa7a7a61986279c6d", "url": "https://huggingface.co/bartowski/Ling-3.0-flash-GGUF/tree/7e1a1243566501d52af98e9aa7a7a61986279c6d", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 2, "status": "measured_file_manifest" }, "GGUF_Q5_K_M": { "bytes": 90661271360, "repo": "bartowski/Ling-3.0-flash-GGUF", "revision": "7e1a1243566501d52af98e9aa7a7a61986279c6d", "url": "https://huggingface.co/bartowski/Ling-3.0-flash-GGUF/tree/7e1a1243566501d52af98e9aa7a7a61986279c6d", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 3, "status": "measured_file_manifest" }, "GGUF_Q8_0": { "bytes": 135633373120, "repo": "bartowski/Ling-3.0-flash-GGUF", "revision": "7e1a1243566501d52af98e9aa7a7a61986279c6d", "url": "https://huggingface.co/bartowski/Ling-3.0-flash-GGUF/tree/7e1a1243566501d52af98e9aa7a7a61986279c6d", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 4, "status": "measured_file_manifest" }, "GGUF_BF16": { "bytes": 255091083264, "repo": "bartowski/Ling-3.0-flash-GGUF", "revision": "7e1a1243566501d52af98e9aa7a7a61986279c6d", "url": "https://huggingface.co/bartowski/Ling-3.0-flash-GGUF/tree/7e1a1243566501d52af98e9aa7a7a61986279c6d", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 7, "status": "measured_file_manifest" }, "MLX_4bit": { "bytes": 70012818813, "repo": "TensorFold/Ling-3.0-flash-MLX-4bit", "revision": "cae62774492b589eea4aab2d6d36fb167a67d90c", "url": "https://huggingface.co/TensorFold/Ling-3.0-flash-MLX-4bit/tree/cae62774492b589eea4aab2d6d36fb167a67d90c", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 14, "status": "measured_file_manifest", "mixed": true }, "MLX_8bit": { "bytes": 132192655480, "repo": "TensorFold/Ling-3.0-flash-MLX-8bit", "revision": "2dfcdab90670d07ed932ae76121bf41a3c6960ce", "url": "https://huggingface.co/TensorFold/Ling-3.0-flash-MLX-8bit/tree/2dfcdab90670d07ed932ae76121bf41a3c6960ce", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 27, "status": "measured_file_manifest", "mixed": true } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [], "notes": [] }, { "slug": "mellum2.1-12b-a2.5b-thinking", "name": "Mellum2.1-12B-A2.5B-Thinking", "hf_repo": "JetBrains/Mellum2.1-12B-A2.5B-Thinking", "hf_revision": "92ddae9fc7665e9f801d141d2e5a6b2caf2460c4", "source_url": "https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking", "org": "JetBrains", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-10-08", "precision": "day", "url": "https://blog.jetbrains.com/ai/2026/10/mellum2-1-gets-to-work-a-fast-open-model-for-coding-agents/", "status": "verified_primary", "note": "Publisher announcement; date also on publisher releases index." }, "params": { "total_b": 12, "active_b": 2.5, "moe": true, "url": "https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking/blob/92ddae9fc7665e9f801d141d2e5a6b2caf2460c4/README.md", "active_url": "https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking/blob/92ddae9fc7665e9f801d141d2e5a6b2caf2460c4/README.md", "scope": "Publisher total/active counts." }, "arch_label": "MoE · sliding-window attn", "context": { "native": 131072, "url": "https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking/blob/92ddae9fc7665e9f801d141d2e5a6b2caf2460c4/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking/blob/92ddae9fc7665e9f801d141d2e5a6b2caf2460c4/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 14336, "win_per_token": 43008, "win_tokens": 1024, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking/blob/92ddae9fc7665e9f801d141d2e5a6b2caf2460c4/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 8071295264, "repo": "JetBrains/Mellum2.1-12B-A2.5B-Thinking-GGUF", "revision": "20b439426f1e997ba575eaccaf4856ac0f76f5d5", "url": "https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking-GGUF/tree/20b439426f1e997ba575eaccaf4856ac0f76f5d5", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 9969220320, "repo": "bartowski/Mellum2.1-12B-A2.5B-Thinking-GGUF", "revision": "355e97042763a75b33efc51efd731ce0149413d4", "url": "https://huggingface.co/bartowski/Mellum2.1-12B-A2.5B-Thinking-GGUF/tree/355e97042763a75b33efc51efd731ce0149413d4", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 12925417760, "repo": "JetBrains/Mellum2.1-12B-A2.5B-Thinking-GGUF", "revision": "20b439426f1e997ba575eaccaf4856ac0f76f5d5", "url": "https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking-GGUF/tree/20b439426f1e997ba575eaccaf4856ac0f76f5d5", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16", "MLX_4bit", "MLX_8bit" ], "community": [ { "platform": "Hugging Face", "date": "2026-10-17", "title": "When is MTP coming?", "url": "https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking/discussions/2" }, { "platform": "Hugging Face", "date": "2026-10-13", "title": "Appreciation and image processing", "url": "https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking/discussions/1" } ], "notes": [] }, { "slug": "mimo-v2.6-distill-qwen-9b", "name": "MiMo-V2.6-Distill-Qwen-9B", "hf_repo": "XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B", "hf_revision": "2367e865d009c13ac81713a2878291d33ab28177", "source_url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B", "org": "XiaomiMiMo", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-09-22", "precision": "day", "url": "https://mimo.mi.com/docs/en-US/news/latest", "status": "verified_primary", "note": "Publisher announcement/update date explicitly says Distill-Qwen-9B was released simultaneously with the series. Earlier Sep21 repository/community timestamps may reflect timezone or earlier availability; exact first public-access instant is not established. Official page is JS-rendered; live indexed primary text was inspected." }, "params": { "total_b": 9, "active_b": 9, "moe": false, "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B/blob/2367e865d009c13ac81713a2878291d33ab28177/README.md", "active_url": null, "scope": "Nominal Qwen3.5-9B-based distilled checkpoint; HF total separately." }, "arch_label": "Dense · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B/blob/2367e865d009c13ac81713a2878291d33ab28177/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B/blob/2367e865d009c13ac81713a2878291d33ab28177/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 32768, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B/blob/2367e865d009c13ac81713a2878291d33ab28177/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 5841049120, "repo": "bartowski/MiMo-V2.6-Distill-Qwen-9B-GGUF", "revision": "4371da10c84fb26da3592d4cf312d24aa82b7b65", "url": "https://huggingface.co/bartowski/MiMo-V2.6-Distill-Qwen-9B-GGUF/tree/4371da10c84fb26da3592d4cf312d24aa82b7b65", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-MiMo-V2.6-Distill-Qwen-9B-bf16.gguf", "bytes": 921704992, "url": "https://huggingface.co/bartowski/MiMo-V2.6-Distill-Qwen-9B-GGUF/blob/4371da10c84fb26da3592d4cf312d24aa82b7b65/mmproj-MiMo-V2.6-Distill-Qwen-9B-bf16.gguf" }, { "path": "mmproj-MiMo-V2.6-Distill-Qwen-9B-f16.gguf", "bytes": 918166048, "url": "https://huggingface.co/bartowski/MiMo-V2.6-Distill-Qwen-9B-GGUF/blob/4371da10c84fb26da3592d4cf312d24aa82b7b65/mmproj-MiMo-V2.6-Distill-Qwen-9B-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 6876124704, "repo": "bartowski/MiMo-V2.6-Distill-Qwen-9B-GGUF", "revision": "4371da10c84fb26da3592d4cf312d24aa82b7b65", "url": "https://huggingface.co/bartowski/MiMo-V2.6-Distill-Qwen-9B-GGUF/tree/4371da10c84fb26da3592d4cf312d24aa82b7b65", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-MiMo-V2.6-Distill-Qwen-9B-bf16.gguf", "bytes": 921704992, "url": "https://huggingface.co/bartowski/MiMo-V2.6-Distill-Qwen-9B-GGUF/blob/4371da10c84fb26da3592d4cf312d24aa82b7b65/mmproj-MiMo-V2.6-Distill-Qwen-9B-bf16.gguf" }, { "path": "mmproj-MiMo-V2.6-Distill-Qwen-9B-f16.gguf", "bytes": 918166048, "url": "https://huggingface.co/bartowski/MiMo-V2.6-Distill-Qwen-9B-GGUF/blob/4371da10c84fb26da3592d4cf312d24aa82b7b65/mmproj-MiMo-V2.6-Distill-Qwen-9B-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 9545979424, "repo": "bartowski/MiMo-V2.6-Distill-Qwen-9B-GGUF", "revision": "4371da10c84fb26da3592d4cf312d24aa82b7b65", "url": "https://huggingface.co/bartowski/MiMo-V2.6-Distill-Qwen-9B-GGUF/tree/4371da10c84fb26da3592d4cf312d24aa82b7b65", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-MiMo-V2.6-Distill-Qwen-9B-bf16.gguf", "bytes": 921704992, "url": "https://huggingface.co/bartowski/MiMo-V2.6-Distill-Qwen-9B-GGUF/blob/4371da10c84fb26da3592d4cf312d24aa82b7b65/mmproj-MiMo-V2.6-Distill-Qwen-9B-bf16.gguf" }, { "path": "mmproj-MiMo-V2.6-Distill-Qwen-9B-f16.gguf", "bytes": 918166048, "url": "https://huggingface.co/bartowski/MiMo-V2.6-Distill-Qwen-9B-GGUF/blob/4371da10c84fb26da3592d4cf312d24aa82b7b65/mmproj-MiMo-V2.6-Distill-Qwen-9B-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 17920693952, "repo": "mradermacher/MiMo-V2.6-Distill-Qwen-9B-GGUF", "revision": "88526399a743fecca877b981758a6fff7f9ed628", "url": "https://huggingface.co/mradermacher/MiMo-V2.6-Distill-Qwen-9B-GGUF/tree/88526399a743fecca877b981758a6fff7f9ed628", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "MiMo-V2.6-Distill-Qwen-9B.mmproj-Q8_0.gguf", "bytes": 624229824, "url": "https://huggingface.co/mradermacher/MiMo-V2.6-Distill-Qwen-9B-GGUF/blob/88526399a743fecca877b981758a6fff7f9ed628/MiMo-V2.6-Distill-Qwen-9B.mmproj-Q8_0.gguf" }, { "path": "MiMo-V2.6-Distill-Qwen-9B.mmproj-f16.gguf", "bytes": 918166080, "url": "https://huggingface.co/mradermacher/MiMo-V2.6-Distill-Qwen-9B-GGUF/blob/88526399a743fecca877b981758a6fff7f9ed628/MiMo-V2.6-Distill-Qwen-9B.mmproj-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 5950219560, "repo": "agnosticeng/MiMo-V2.6-Distill-Qwen-9B-4bit", "revision": "86998933799bdba2eb380c31d12810e9c289e724", "url": "https://huggingface.co/agnosticeng/MiMo-V2.6-Distill-Qwen-9B-4bit/tree/86998933799bdba2eb380c31d12810e9c289e724", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 2, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 10426590911, "repo": "nativ-community/MiMo-V2.6-Distill-Qwen-9B-MLX-8bit", "revision": "7319fec70dd29d6a5f89e5eb9ad8820152842ad4", "url": "https://huggingface.co/nativ-community/MiMo-V2.6-Distill-Qwen-9B-MLX-8bit/tree/7319fec70dd29d6a5f89e5eb9ad8820152842ad4", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 2, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-10-16", "title": "GGUF Q8_0 Forensic Benchmark Audit (llama.cpp / Ollama v0.40.4, RTX 5080): 12x Faster than Base Qwen3.5-9B at Equal 96% GSM8K Ceiling + Thermal Sweep (T=0.0–0.3)", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B/discussions/15" }, { "platform": "Hugging Face", "date": "2026-10-09", "title": "GGUF/llama.cpp results: YaRN 512K works: 40/40 needle retrieval at 512K, zero short-task quality loss, ~660 tok/s prefill (llama.cpp, RX 7900 XTX)", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B/discussions/13" } ], "notes": [ "The verified September 22 publisher announcement supersedes the earlier unresolved-date research note; the release_date qualification preserves the September 21 timestamp discrepancy." ] }, { "slug": "mimo-v2.6-flash", "name": "MiMo-V2.6-Flash", "hf_repo": "XiaomiMiMo/MiMo-V2.6-Flash-RL", "hf_revision": "dcd3fd83fb181622fbc606fbb1709901982a62c2", "source_url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL", "org": "XiaomiMiMo", "status": "superseded", "catalog_status": "superseded_2026_checkpoint", "release_2026_verified": true, "default_visible": false, "release": { "label": "2026-09-22", "precision": "day", "url": "https://mimo.mi.com/docs/en-US/news/latest", "status": "verified_primary", "note": "Publisher release docs date Sep22; original X posts Sep21 can reflect timezone. Keep both original series and later MOPD dates." }, "params": { "total_b": 309, "active_b": 15, "moe": true, "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL/raw/dcd3fd83fb181622fbc606fbb1709901982a62c2/README.md", "active_url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL/raw/dcd3fd83fb181622fbc606fbb1709901982a62c2/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL/raw/dcd3fd83fb181622fbc606fbb1709901982a62c2/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL/blob/dcd3fd83fb181622fbc606fbb1709901982a62c2/LICENSE" }, "quality": { "rank": null, "aa": { "value": 37.8843590141754, "variant": "MiMo-V2.6-Flash", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/mimo-v2-6-flash", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1455.2141256063, "lower": 1448.7630895366, "upper": 1461.665161676, "votes": 10917, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL/raw/dcd3fd83fb181622fbc606fbb1709901982a62c2/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "MLX_4bit": { "bytes": 168999005012, "repo": "nativ-community/MiMo-V2.6-Flash-RL-MLX-4bit", "revision": "3fb80d49283ba54671ad62baa1f595beb8b8a9e9", "url": "https://huggingface.co/nativ-community/MiMo-V2.6-Flash-RL-MLX-4bit/tree/3fb80d49283ba54671ad62baa1f595beb8b8a9e9", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 36, "status": "measured_file_manifest", "mixed": true } }, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_Q8_0", "GGUF_FP16", "GGUF_BF16", "MLX_8bit" ], "community": [], "notes": [ "RL and MOPD are distinct checkpoints. API switched to MOPD on Sep25; hosted AA/Arena scores are not proven identical to this local checkpoint." ], "superseded_by": "mimo-v2.6-flash-mopd" }, { "slug": "mimo-v2.6-pro", "name": "MiMo-V2.6-Pro", "hf_repo": "XiaomiMiMo/MiMo-V2.6-Pro-RL", "hf_revision": "b61f21852680ee02a29e946e18cf8697f4844061", "source_url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL", "org": "XiaomiMiMo", "status": "superseded", "catalog_status": "superseded_2026_checkpoint", "release_2026_verified": true, "default_visible": false, "release": { "label": "2026-09-22", "precision": "day", "url": "https://mimo.mi.com/docs/en-US/news/latest", "status": "verified_primary", "note": "Publisher release docs date Sep22; original X posts Sep21 can reflect timezone. Keep both original series and later MOPD dates." }, "params": { "total_b": 1020, "active_b": 42, "moe": true, "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL/raw/b61f21852680ee02a29e946e18cf8697f4844061/README.md", "active_url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL/raw/b61f21852680ee02a29e946e18cf8697f4844061/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL/raw/b61f21852680ee02a29e946e18cf8697f4844061/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL/blob/b61f21852680ee02a29e946e18cf8697f4844061/LICENSE" }, "quality": { "rank": null, "aa": { "value": 46.3242065310383, "variant": "MiMo-V2.6-Pro", "estimated": false, "url": "https://artificialanalysis.ai/leaderboards/models", "page_url": "https://artificialanalysis.ai/models/mimo-v2-6-pro", "observed_at": "2026-10-17", "status": "benchmark_publisher_reported" }, "arena": { "value": 1490.6785014822, "lower": 1481.349339658, "upper": 1500.0076633063, "votes": 4075, "date": "2026-10-15", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/main/text/latest-00000-of-00001.parquet", "page_url": "https://arena.ai/leaderboard/chat/text", "mapping": "Exact listed model name; hosted reasoning/quantization may differ from local checkpoint." } }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL/raw/b61f21852680ee02a29e946e18cf8697f4844061/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": {}, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_Q8_0", "GGUF_FP16", "GGUF_BF16", "MLX_4bit", "MLX_8bit" ], "community": [], "notes": [ "RL and MOPD are distinct checkpoints. API switched to MOPD on Sep25; hosted AA/Arena scores are not proven identical to this local checkpoint." ], "superseded_by": "mimo-v2.6-pro-mopd" }, { "slug": "minicpm-v-4.7-35b-a3b", "name": "MiniCPM-V 4.7 35B-A3B", "hf_repo": "openbmb/MiniCPM-V-4.7-35B-A3B", "hf_revision": "0bbda78419a0b505a58234574ff40b1c344a86fc", "source_url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B", "org": "openbmb", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-10-12", "precision": "day", "url": "https://github.com/OpenBMB/MiniCPM-V", "status": "verified_primary", "note": "Official latest-updates release entry; not October 7 repository creation." }, "params": { "total_b": 35, "active_b": 3, "moe": true, "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B/blob/0bbda78419a0b505a58234574ff40b1c344a86fc/README.md", "active_url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B/blob/0bbda78419a0b505a58234574ff40b1c344a86fc/README.md", "scope": "Publisher nominal total/active count." }, "arch_label": "MoE · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B/blob/0bbda78419a0b505a58234574ff40b1c344a86fc/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B/blob/0bbda78419a0b505a58234574ff40b1c344a86fc/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 20480, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B/blob/0bbda78419a0b505a58234574ff40b1c344a86fc/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 21166255008, "repo": "openbmb/MiniCPM-V-4.7-35B-A3B-gguf", "revision": "77ec40de8445999bf3076aeff9ff45d40a2713cc", "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B-gguf/tree/77ec40de8445999bf3076aeff9ff45d40a2713cc", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-model-f16-4x.gguf", "bytes": 910491648, "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B-gguf/blob/77ec40de8445999bf3076aeff9ff45d40a2713cc/mmproj-model-f16-4x.gguf" }, { "path": "mmproj-model-f16.gguf", "bytes": 1119575840, "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B-gguf/blob/77ec40de8445999bf3076aeff9ff45d40a2713cc/mmproj-model-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 24728583072, "repo": "openbmb/MiniCPM-V-4.7-35B-A3B-gguf", "revision": "77ec40de8445999bf3076aeff9ff45d40a2713cc", "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B-gguf/tree/77ec40de8445999bf3076aeff9ff45d40a2713cc", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-model-f16-4x.gguf", "bytes": 910491648, "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B-gguf/blob/77ec40de8445999bf3076aeff9ff45d40a2713cc/mmproj-model-f16-4x.gguf" }, { "path": "mmproj-model-f16.gguf", "bytes": 1119575840, "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B-gguf/blob/77ec40de8445999bf3076aeff9ff45d40a2713cc/mmproj-model-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 36902368928, "repo": "openbmb/MiniCPM-V-4.7-35B-A3B-gguf", "revision": "77ec40de8445999bf3076aeff9ff45d40a2713cc", "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B-gguf/tree/77ec40de8445999bf3076aeff9ff45d40a2713cc", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-model-f16-4x.gguf", "bytes": 910491648, "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B-gguf/blob/77ec40de8445999bf3076aeff9ff45d40a2713cc/mmproj-model-f16-4x.gguf" }, { "path": "mmproj-model-f16.gguf", "bytes": 1119575840, "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B-gguf/blob/77ec40de8445999bf3076aeff9ff45d40a2713cc/mmproj-model-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 69375191008, "repo": "openbmb/MiniCPM-V-4.7-35B-A3B-gguf", "revision": "77ec40de8445999bf3076aeff9ff45d40a2713cc", "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B-gguf/tree/77ec40de8445999bf3076aeff9ff45d40a2713cc", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-model-f16-4x.gguf", "bytes": 910491648, "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B-gguf/blob/77ec40de8445999bf3076aeff9ff45d40a2713cc/mmproj-model-f16-4x.gguf" }, { "path": "mmproj-model-f16.gguf", "bytes": 1119575840, "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B-gguf/blob/77ec40de8445999bf3076aeff9ff45d40a2713cc/mmproj-model-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 23528366118, "repo": "mlx-community/MiniCPM-V-4.7-35B-A3B-4bit", "revision": "0e3b706c1599ad95294122c67c77ce662de18794", "url": "https://huggingface.co/mlx-community/MiniCPM-V-4.7-35B-A3B-4bit/tree/0e3b706c1599ad95294122c67c77ce662de18794", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 5, "status": "verified_hf_file_sizes", "mixed": true, "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 39833722632, "repo": "mlx-community/MiniCPM-V-4.7-35B-A3B-8bit", "revision": "f7d166710f377cff3981ecab850ba83d2aef07b7", "url": "https://huggingface.co/mlx-community/MiniCPM-V-4.7-35B-A3B-8bit/tree/f7d166710f377cff3981ecab850ba83d2aef07b7", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 8, "status": "verified_hf_file_sizes", "mixed": true, "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-10-15", "title": "🚀 37.5 t/s on RTX 2060 Super (8GB): Optimizing MiniCPM-V 4.7 35B-A3B with DSpark & CPU-MoE", "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B/discussions/13" }, { "platform": "Hugging Face", "date": "2026-10-14", "title": "MiniCPM-V 4.7 vs Qwen3.6-35B-A3B in 5 physics simulation challenges", "url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B/discussions/12" } ], "notes": [] }, { "slug": "minimax-m2.7", "name": "MiniMax-M2.7", "hf_repo": "MiniMaxAI/MiniMax-M2.7", "hf_revision": "d494266a4affc0d2995ba1fa35c8481cbd84294b", "source_url": "https://huggingface.co/MiniMaxAI/MiniMax-M2.7", "org": "MiniMaxAI", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026", "precision": "year", "url": "https://www.minimax.io/news/minimax-m27-en", "status": "verified_publisher_announcement_year", "note": "API/model announcement March18; public weight date likely April12 from community, not verified primary exact day." }, "params": { "total_b": null, "active_b": null, "moe": true, "url": null, "active_url": null, "scope": null }, "arch_label": "MoE", "context": { "native": 204800, "url": "https://huggingface.co/MiniMaxAI/MiniMax-M2.7/raw/d494266a4affc0d2995ba1fa35c8481cbd84294b/config.json", "extended": null }, "license": { "value": "other", "url": "https://github.com/MiniMax-AI/MiniMax-M2.7/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 253952, "win_per_token": null, "win_tokens": null, "win_max": 0, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/MiniMaxAI/MiniMax-M2.7/raw/d494266a4affc0d2995ba1fa35c8481cbd84294b/config.json", "dtype": "BF16", "note": "BF16/FP16 KV estimate for full-attention layers only, batch=1. Sliding-window terms and fixed recurrent state are separate; excludes alignment, graphs, buffers, vision and MTP/drafters." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 138805347296, "repo": "bartowski/MiniMaxAI_MiniMax-M2.7-GGUF", "revision": "2b13ec99437c3ddb36f692a3a90be57ed3ba43af", "url": "https://huggingface.co/bartowski/MiniMaxAI_MiniMax-M2.7-GGUF/tree/2b13ec99437c3ddb36f692a3a90be57ed3ba43af", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 4, "status": "measured_file_manifest" }, "GGUF_Q5_K_M": { "bytes": 162668299360, "repo": "bartowski/MiniMaxAI_MiniMax-M2.7-GGUF", "revision": "2b13ec99437c3ddb36f692a3a90be57ed3ba43af", "url": "https://huggingface.co/bartowski/MiniMaxAI_MiniMax-M2.7-GGUF/tree/2b13ec99437c3ddb36f692a3a90be57ed3ba43af", "precision": "GGUF Q5_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 5, "status": "measured_file_manifest" }, "GGUF_Q8_0": { "bytes": 243136872992, "repo": "unsloth/MiniMax-M2.7-GGUF", "revision": "d2a05ccf69491b03db0cc40b335aec14bdaf7198", "url": "https://huggingface.co/unsloth/MiniMax-M2.7-GGUF/tree/d2a05ccf69491b03db0cc40b335aec14bdaf7198", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 6, "status": "measured_file_manifest" }, "GGUF_BF16": { "bytes": 457487024928, "repo": "unsloth/MiniMax-M2.7-GGUF", "revision": "d2a05ccf69491b03db0cc40b335aec14bdaf7198", "url": "https://huggingface.co/unsloth/MiniMax-M2.7-GGUF/tree/d2a05ccf69491b03db0cc40b335aec14bdaf7198", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 10, "status": "measured_file_manifest" }, "MLX_4bit": { "bytes": 128666887980, "repo": "mlx-community/MiniMax-M2.7-4bit", "revision": "66d2e5cb7c5cda05251b4625c504af4b034df7ff", "url": "https://huggingface.co/mlx-community/MiniMax-M2.7-4bit/tree/66d2e5cb7c5cda05251b4625c504af4b034df7ff", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 27, "status": "measured_file_manifest", "mixed": true }, "MLX_8bit": { "bytes": 242986969522, "repo": "mlx-community/MiniMax-M2.7-8bit", "revision": "a4236145d7deca7e49f098b1ff9de0ee066d58ed", "url": "https://huggingface.co/mlx-community/MiniMax-M2.7-8bit/tree/a4236145d7deca7e49f098b1ff9de0ee066d58ed", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 47, "status": "measured_file_manifest", "mixed": true } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [], "notes": [] }, { "slug": "nvidia-nemotron-3-super-120b-a12b", "name": "NVIDIA Nemotron 3 Super 120B-A12B", "hf_repo": "nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16", "hf_revision": "2dc98e2afe4face0e4ce40972a915c45368bd34a", "source_url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16", "org": "nvidia", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-03-11", "precision": "day", "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16/raw/main/README.md", "status": "verified_primary", "note": null }, "params": { "total_b": 120, "active_b": 12, "moe": true, "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16/raw/2dc98e2afe4face0e4ce40972a915c45368bd34a/README.md", "active_url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16/raw/2dc98e2afe4face0e4ce40972a915c45368bd34a/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 262144, "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16/raw/2dc98e2afe4face0e4ce40972a915c45368bd34a/config.json", "extended": null }, "license": { "value": "nvidia-nemotron-open-model-license", "url": "https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-nemotron-open-model-license/" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 8192, "win_per_token": null, "win_tokens": null, "win_max": 0, "fixed": null, "unknown": [ "sliding_window", "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16/raw/2dc98e2afe4face0e4ce40972a915c45368bd34a/config.json", "dtype": "BF16", "note": "Growing attention KV only. Mamba/recurrent/convolution states are extra and backend-specific. FFN/MoE blocks do not each add KV." }, "artifacts": { "GGUF_Q8_0": { "bytes": 128472991680, "repo": "unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF", "revision": "036038fb30334a2d56a146c6f0d4871ab5edccbb", "url": "https://huggingface.co/unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF/tree/036038fb30334a2d56a146c6f0d4871ab5edccbb", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 4, "status": "measured_file_manifest" }, "GGUF_BF16": { "bytes": 241518659424, "repo": "unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF", "revision": "036038fb30334a2d56a146c6f0d4871ab5edccbb", "url": "https://huggingface.co/unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF/tree/036038fb30334a2d56a146c6f0d4871ab5edccbb", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 5, "status": "measured_file_manifest" }, "MLX_4bit": { "bytes": 68000941086, "repo": "mlx-community/NVIDIA-Nemotron-3-Super-120B-A12B-4bit", "revision": "46d2ec99579734e0da13b6189bb7ea500b7a1e00", "url": "https://huggingface.co/mlx-community/NVIDIA-Nemotron-3-Super-120B-A12B-4bit/tree/46d2ec99579734e0da13b6189bb7ea500b7a1e00", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 14, "status": "measured_file_manifest" }, "MLX_8bit": { "bytes": 128291964658, "repo": "mlx-community/NVIDIA-Nemotron-3-Super-120B-A12B-8bit", "revision": "c505de746dab3eda1f557f01fb94b1a2e91ba0e5", "url": "https://huggingface.co/mlx-community/NVIDIA-Nemotron-3-Super-120B-A12B-8bit/tree/c505de746dab3eda1f557f01fb94b1a2e91ba0e5", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 27, "status": "measured_file_manifest" } }, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_FP16" ], "community": [], "notes": [ "Card advertises 1M context while current config max_position_embeddings is 262144." ] }, { "slug": "nanbeige4.2-3b", "name": "Nanbeige4.2-3B", "hf_repo": "Nanbeige/Nanbeige4.2-3B", "hf_revision": "b82e54bd609793562a75cbf9337970a93369eab5", "source_url": "https://huggingface.co/Nanbeige/Nanbeige4.2-3B", "org": "Nanbeige", "status": "candidate", "catalog_status": "candidate_release_date_unresolved", "release_2026_verified": false, "default_visible": false, "release": { "label": null, "precision": "unknown", "url": null, "status": "unknown", "note": null }, "params": { "total_b": 4, "active_b": 4, "moe": false, "url": "https://huggingface.co/Nanbeige/Nanbeige4.2-3B/blob/b82e54bd609793562a75cbf9337970a93369eab5/README.md", "active_url": null, "scope": "Benchmark table states 4B total / 3B non-embedding." }, "arch_label": "Dense", "context": { "native": 262144, "url": "https://huggingface.co/Nanbeige/Nanbeige4.2-3B/blob/b82e54bd609793562a75cbf9337970a93369eab5/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/Nanbeige/Nanbeige4.2-3B/blob/b82e54bd609793562a75cbf9337970a93369eab5/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/Nanbeige/Nanbeige4.2-3B/blob/b82e54bd609793562a75cbf9337970a93369eab5/config.json", "dtype": "FP16", "note": "Architecture needs backend-specific treatment: shared K/V or cross-layer cache, looped layers, MoVA, MLA/KDA or diffusion. No naive all-layers GQA estimate is supplied." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 2574808384, "repo": "mradermacher/Nanbeige4.2-3B-GGUF", "revision": "c2c32add1869793644a09f12cfa0a7392f170f3d", "url": "https://huggingface.co/mradermacher/Nanbeige4.2-3B-GGUF/tree/c2c32add1869793644a09f12cfa0a7392f170f3d", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 2986997056, "repo": "mradermacher/Nanbeige4.2-3B-GGUF", "revision": "c2c32add1869793644a09f12cfa0a7392f170f3d", "url": "https://huggingface.co/mradermacher/Nanbeige4.2-3B-GGUF/tree/c2c32add1869793644a09f12cfa0a7392f170f3d", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 4434787648, "repo": "mradermacher/Nanbeige4.2-3B-GGUF", "revision": "c2c32add1869793644a09f12cfa0a7392f170f3d", "url": "https://huggingface.co/mradermacher/Nanbeige4.2-3B-GGUF/tree/c2c32add1869793644a09f12cfa0a7392f170f3d", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 8343846208, "repo": "mradermacher/Nanbeige4.2-3B-GGUF", "revision": "c2c32add1869793644a09f12cfa0a7392f170f3d", "url": "https://huggingface.co/mradermacher/Nanbeige4.2-3B-GGUF/tree/c2c32add1869793644a09f12cfa0a7392f170f3d", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 2345769894, "repo": "MercuriusDream/Nanbeige4.2-3B-mlx-4bit", "revision": "c21f850c4096b67a1b53a2c8eb4f8c8652f89375", "url": "https://huggingface.co/MercuriusDream/Nanbeige4.2-3B-mlx-4bit/tree/c21f850c4096b67a1b53a2c8eb4f8c8652f89375", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 4430601410, "repo": "MercuriusDream/Nanbeige4.2-3B-mlx-8bit", "revision": "7737363155c8a6de6feb7ef3b66e796a2b002dcc", "url": "https://huggingface.co/MercuriusDream/Nanbeige4.2-3B-mlx-8bit/tree/7737363155c8a6de6feb7ef3b66e796a2b002dcc", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-09-10", "title": "Q4_K_M slower than Q8_0", "url": "https://huggingface.co/Nanbeige/Nanbeige4.2-3B/discussions/35" }, { "platform": "Hugging Face", "date": "2026-08-13", "title": "Why is it much slower and much more memory hungry than slightly larger models such as Qwen3.5-4B?", "url": "https://huggingface.co/Nanbeige/Nanbeige4.2-3B/discussions/31" } ], "notes": [ "Exact public release day not verified from a primary announcement; repository creation and commit dates are deliberately not substituted." ] }, { "slug": "ornith-1.5-9b", "name": "Ornith-1.5-9B", "hf_repo": "ornith-ai/Ornith-1.5-9B", "hf_revision": "489cb97981b8654bcfcf30ce1f94ed1b62e07b53", "source_url": "https://huggingface.co/ornith-ai/Ornith-1.5-9B", "org": "ornith-ai", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-08", "precision": "month", "url": "https://ornith.ai/ornith_1_5.html", "status": "verified_primary", "note": "Publisher blog states August 2026 only; exact day unknown." }, "params": { "total_b": 9, "active_b": 9, "moe": false, "url": "https://huggingface.co/ornith-ai/Ornith-1.5-9B/blob/489cb97981b8654bcfcf30ce1f94ed1b62e07b53/README.md", "active_url": null, "scope": "Dense Qwen3.5-based derivative; nominal publisher size." }, "arch_label": "Dense · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/ornith-ai/Ornith-1.5-9B/blob/489cb97981b8654bcfcf30ce1f94ed1b62e07b53/config.json", "extended": null }, "license": { "value": "mit", "url": "https://huggingface.co/ornith-ai/Ornith-1.5-9B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 32768, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/ornith-ai/Ornith-1.5-9B/blob/489cb97981b8654bcfcf30ce1f94ed1b62e07b53/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 5780090816, "repo": "ornith-ai/Ornith-1.5-9B-GGUF", "revision": "abdd624b12ebf020b767fff532ff44fe552b28c3", "url": "https://huggingface.co/ornith-ai/Ornith-1.5-9B-GGUF/tree/abdd624b12ebf020b767fff532ff44fe552b28c3", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-Ornith-1.5-9B-BF16.gguf", "bytes": 921704672, "url": "https://huggingface.co/ornith-ai/Ornith-1.5-9B-GGUF/blob/abdd624b12ebf020b767fff532ff44fe552b28c3/mmproj-Ornith-1.5-9B-BF16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 6642544576, "repo": "ornith-ai/Ornith-1.5-9B-GGUF", "revision": "abdd624b12ebf020b767fff532ff44fe552b28c3", "url": "https://huggingface.co/ornith-ai/Ornith-1.5-9B-GGUF/tree/abdd624b12ebf020b767fff532ff44fe552b28c3", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-Ornith-1.5-9B-BF16.gguf", "bytes": 921704672, "url": "https://huggingface.co/ornith-ai/Ornith-1.5-9B-GGUF/blob/abdd624b12ebf020b767fff532ff44fe552b28c3/mmproj-Ornith-1.5-9B-BF16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 9786060384, "repo": "ornith-ai/Ornith-1.5-9B-GGUF", "revision": "abdd624b12ebf020b767fff532ff44fe552b28c3", "url": "https://huggingface.co/ornith-ai/Ornith-1.5-9B-GGUF/tree/abdd624b12ebf020b767fff532ff44fe552b28c3", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-Ornith-1.5-9B-BF16.gguf", "bytes": 921704672, "url": "https://huggingface.co/ornith-ai/Ornith-1.5-9B-GGUF/blob/abdd624b12ebf020b767fff532ff44fe552b28c3/mmproj-Ornith-1.5-9B-BF16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 17920697376, "repo": "mradermacher/Ornith-1.5-9B-GGUF", "revision": "39c0ed89831977862629347f2ac18cc3b0347785", "url": "https://huggingface.co/mradermacher/Ornith-1.5-9B-GGUF/tree/39c0ed89831977862629347f2ac18cc3b0347785", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "Ornith-1.5-9B.mmproj-Q8_0.gguf", "bytes": 624229472, "url": "https://huggingface.co/mradermacher/Ornith-1.5-9B-GGUF/blob/39c0ed89831977862629347f2ac18cc3b0347785/Ornith-1.5-9B.mmproj-Q8_0.gguf" }, { "path": "Ornith-1.5-9B.mmproj-f16.gguf", "bytes": 918165728, "url": "https://huggingface.co/mradermacher/Ornith-1.5-9B-GGUF/blob/39c0ed89831977862629347f2ac18cc3b0347785/Ornith-1.5-9B.mmproj-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 5038161163, "repo": "ornith-ai/Ornith-1.5-9B-MLX-4bit", "revision": "a48173b246ac705be75c05bedf1a0666db522d53", "url": "https://huggingface.co/ornith-ai/Ornith-1.5-9B-MLX-4bit/tree/a48173b246ac705be75c05bedf1a0666db522d53", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 9514532079, "repo": "ornith-ai/Ornith-1.5-9B-MLX-8bit", "revision": "b4b70543d60c81e7c418a001b74cd4352212bd44", "url": "https://huggingface.co/ornith-ai/Ornith-1.5-9B-MLX-8bit/tree/b4b70543d60c81e7c418a001b74cd4352212bd44", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 2, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-09-16", "title": "The reasoning speed of this model is really fast, but the quality is just average.", "url": "https://huggingface.co/ornith-ai/Ornith-1.5-9B/discussions/18" }, { "platform": "Hugging Face", "date": "2026-10-11", "title": "Model thinks very, VERY long", "url": "https://huggingface.co/ornith-ai/Ornith-1.5-9B/discussions/20" }, { "platform": "Reddit", "date": "2026-09-15", "title": null, "url": "https://www.reddit.com/r/LocalLLaMA/comments/1wha6sv/i_benchmarked_ifmk2horizon7b_on_16gb_vram/" } ], "notes": [ "Exact public release day not verified from a primary announcement; repository creation and commit dates are deliberately not substituted." ] }, { "slug": "laguna-s-2-1", "name": "Poolside Laguna S 2.1", "hf_repo": "poolside/Laguna-S-2.1", "hf_revision": "0f573140834b11cfac0c2af97a101a7a69a13e22", "source_url": "https://huggingface.co/poolside/Laguna-S-2.1", "org": "poolside", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-07-21", "precision": "day", "url": "https://poolside.ai/blog/introducing-laguna-s-2-1", "status": "verified_primary", "note": null }, "params": { "total_b": 118, "active_b": 8, "moe": true, "url": "https://huggingface.co/poolside/Laguna-S-2.1/raw/0f573140834b11cfac0c2af97a101a7a69a13e22/README.md", "active_url": "https://huggingface.co/poolside/Laguna-S-2.1/raw/0f573140834b11cfac0c2af97a101a7a69a13e22/README.md", "scope": null }, "arch_label": "MoE · sliding-window attn", "context": { "native": 1048576, "url": "https://huggingface.co/poolside/Laguna-S-2.1/raw/0f573140834b11cfac0c2af97a101a7a69a13e22/config.json", "extended": null }, "license": { "value": "openmdw-1.1", "url": "https://huggingface.co/poolside/Laguna-S-2.1/blob/0f573140834b11cfac0c2af97a101a7a69a13e22/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 49152, "win_per_token": null, "win_tokens": null, "win_max": 75497472, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/poolside/Laguna-S-2.1/raw/0f573140834b11cfac0c2af97a101a7a69a13e22/config.json", "dtype": "BF16", "note": "BF16/FP16 KV estimate for full-attention layers only, batch=1. Sliding-window terms and fixed recurrent state are separate; excludes alignment, graphs, buffers, vision and MTP/drafters." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 96031829760, "repo": "poolside/Laguna-S-2.1-GGUF", "revision": "19036076775e9ba6758595f78af99cb976ef8ff0", "url": "https://huggingface.co/poolside/Laguna-S-2.1-GGUF/tree/19036076775e9ba6758595f78af99cb976ef8ff0", "precision": "GGUF Q4_K_M", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 1, "status": "measured_file_manifest" }, "GGUF_Q8_0": { "bytes": 128750823168, "repo": "poolside/Laguna-S-2.1-GGUF", "revision": "19036076775e9ba6758595f78af99cb976ef8ff0", "url": "https://huggingface.co/poolside/Laguna-S-2.1-GGUF/tree/19036076775e9ba6758595f78af99cb976ef8ff0", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 1, "status": "measured_file_manifest" }, "GGUF_FP16": { "bytes": 235202258240, "repo": "poolside/Laguna-S-2.1-GGUF", "revision": "19036076775e9ba6758595f78af99cb976ef8ff0", "url": "https://huggingface.co/poolside/Laguna-S-2.1-GGUF/tree/19036076775e9ba6758595f78af99cb976ef8ff0", "precision": "GGUF FP16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 1, "status": "measured_file_manifest" }, "GGUF_BF16": { "bytes": 235202259648, "repo": "unsloth/Laguna-S-2.1-GGUF", "revision": "750f92f90cf54159c4d7a610cb7b3e74498e75c6", "url": "https://huggingface.co/unsloth/Laguna-S-2.1-GGUF/tree/750f92f90cf54159c4d7a610cb7b3e74498e75c6", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 5, "status": "measured_file_manifest" }, "MLX_4bit": { "bytes": 66147791039, "repo": "TensorFold/Laguna-S-2.1-MLX-4bit", "revision": "df803a073dfef42c3e90cd795ad490aef65846b9", "url": "https://huggingface.co/TensorFold/Laguna-S-2.1-MLX-4bit/tree/df803a073dfef42c3e90cd795ad490aef65846b9", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 13, "status": "measured_file_manifest", "mixed": true }, "MLX_8bit": { "bytes": 124910138151, "repo": "TensorFold/Laguna-S-2.1-MLX-8bit", "revision": "872720b3a6ff831dee0e228f50d317107b449c3f", "url": "https://huggingface.co/TensorFold/Laguna-S-2.1-MLX-8bit/tree/872720b3a6ff831dee0e228f50d317107b449c3f", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 24, "status": "measured_file_manifest", "mixed": true } }, "artifacts_missing": [ "GGUF_Q5_K_M" ], "community": [], "notes": [] }, { "slug": "qwen3.5-0.8b", "name": "Qwen3.5-0.8B", "hf_repo": "Qwen/Qwen3.5-0.8B", "hf_revision": "2fc06364715b967f1860aea9cf38778875588b17", "source_url": "https://huggingface.co/Qwen/Qwen3.5-0.8B", "org": "Qwen", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-03-02", "precision": "day", "url": "https://github.com/QwenLM/Qwen3.8/blob/main/README.md", "status": "verified_primary", "note": "Official dated release-news entry states public model availability." }, "params": { "total_b": 0.8, "active_b": 0.8, "moe": false, "url": "https://huggingface.co/Qwen/Qwen3.5-0.8B/blob/2fc06364715b967f1860aea9cf38778875588b17/README.md", "active_url": null, "scope": "Publisher nominal language-model size; HF total includes stored multimodal/MTP tensors as applicable." }, "arch_label": "Dense · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/Qwen/Qwen3.5-0.8B/blob/2fc06364715b967f1860aea9cf38778875588b17/config.json", "extended": 1010000 }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/Qwen/Qwen3.5-0.8B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 12288, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3.5-0.8B/blob/2fc06364715b967f1860aea9cf38778875588b17/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 532517120, "repo": "unsloth/Qwen3.5-0.8B-GGUF", "revision": "6ab461498e2023f6e3c1baea90a8f0fe38ab64d0", "url": "https://huggingface.co/unsloth/Qwen3.5-0.8B-GGUF/tree/6ab461498e2023f6e3c1baea90a8f0fe38ab64d0", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 207346528, "url": "https://huggingface.co/unsloth/Qwen3.5-0.8B-GGUF/blob/6ab461498e2023f6e3c1baea90a8f0fe38ab64d0/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 204987232, "url": "https://huggingface.co/unsloth/Qwen3.5-0.8B-GGUF/blob/6ab461498e2023f6e3c1baea90a8f0fe38ab64d0/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 402381664, "url": "https://huggingface.co/unsloth/Qwen3.5-0.8B-GGUF/blob/6ab461498e2023f6e3c1baea90a8f0fe38ab64d0/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 590057728, "repo": "unsloth/Qwen3.5-0.8B-GGUF", "revision": "6ab461498e2023f6e3c1baea90a8f0fe38ab64d0", "url": "https://huggingface.co/unsloth/Qwen3.5-0.8B-GGUF/tree/6ab461498e2023f6e3c1baea90a8f0fe38ab64d0", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 207346528, "url": "https://huggingface.co/unsloth/Qwen3.5-0.8B-GGUF/blob/6ab461498e2023f6e3c1baea90a8f0fe38ab64d0/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 204987232, "url": "https://huggingface.co/unsloth/Qwen3.5-0.8B-GGUF/blob/6ab461498e2023f6e3c1baea90a8f0fe38ab64d0/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 402381664, "url": "https://huggingface.co/unsloth/Qwen3.5-0.8B-GGUF/blob/6ab461498e2023f6e3c1baea90a8f0fe38ab64d0/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 811843840, "repo": "unsloth/Qwen3.5-0.8B-GGUF", "revision": "6ab461498e2023f6e3c1baea90a8f0fe38ab64d0", "url": "https://huggingface.co/unsloth/Qwen3.5-0.8B-GGUF/tree/6ab461498e2023f6e3c1baea90a8f0fe38ab64d0", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 207346528, "url": "https://huggingface.co/unsloth/Qwen3.5-0.8B-GGUF/blob/6ab461498e2023f6e3c1baea90a8f0fe38ab64d0/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 204987232, "url": "https://huggingface.co/unsloth/Qwen3.5-0.8B-GGUF/blob/6ab461498e2023f6e3c1baea90a8f0fe38ab64d0/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 402381664, "url": "https://huggingface.co/unsloth/Qwen3.5-0.8B-GGUF/blob/6ab461498e2023f6e3c1baea90a8f0fe38ab64d0/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 1516745024, "repo": "mradermacher/Qwen3.5-0.8B-GGUF", "revision": "9d48fdbc0d8f133716da87ec1d904e5d2c7175a6", "url": "https://huggingface.co/mradermacher/Qwen3.5-0.8B-GGUF/tree/9d48fdbc0d8f133716da87ec1d904e5d2c7175a6", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "Qwen3.5-0.8B.mmproj-Q8_0.gguf", "bytes": 115923680, "url": "https://huggingface.co/mradermacher/Qwen3.5-0.8B-GGUF/blob/9d48fdbc0d8f133716da87ec1d904e5d2c7175a6/Qwen3.5-0.8B.mmproj-Q8_0.gguf" }, { "path": "Qwen3.5-0.8B.mmproj-f16.gguf", "bytes": 204987104, "url": "https://huggingface.co/mradermacher/Qwen3.5-0.8B-GGUF/blob/9d48fdbc0d8f133716da87ec1d904e5d2c7175a6/Qwen3.5-0.8B.mmproj-f16.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 625229487, "repo": "mlx-community/Qwen3.5-0.8B-4bit", "revision": "da28692b5f139cb0ec58a356b437486b7dac7462", "url": "https://huggingface.co/mlx-community/Qwen3.5-0.8B-4bit/tree/da28692b5f139cb0ec58a356b437486b7dac7462", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 1001176958, "repo": "mlx-community/Qwen3.5-0.8B-8bit", "revision": "87e768fbfa03994095f3d14527c80c5ae70c5758", "url": "https://huggingface.co/mlx-community/Qwen3.5-0.8B-8bit/tree/87e768fbfa03994095f3d14527c80c5ae70c5758", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-04-09", "title": "0.8B with mtp is slower than without mtp", "url": "https://huggingface.co/Qwen/Qwen3.5-0.8B/discussions/15" }, { "platform": "Hugging Face", "date": "2026-04-03", "title": "When vllm deployment is started using mtp config, an error will be reported. Please help！", "url": "https://huggingface.co/Qwen/Qwen3.5-0.8B/discussions/13" } ], "notes": [] }, { "slug": "qwen3.5-27b", "name": "Qwen3.5-27B", "hf_repo": "Qwen/Qwen3.5-27B", "hf_revision": "fc05daec18b0a78c049392ed2e771dde82bdf654", "source_url": "https://huggingface.co/Qwen/Qwen3.5-27B", "org": "Qwen", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-02-24", "precision": "day", "url": "https://github.com/QwenLM/Qwen3.8/blob/main/README.md", "status": "verified_primary", "note": "Official dated release-news entry states public model availability." }, "params": { "total_b": 27, "active_b": 27, "moe": false, "url": "https://huggingface.co/Qwen/Qwen3.5-27B/blob/fc05daec18b0a78c049392ed2e771dde82bdf654/README.md", "active_url": null, "scope": "Publisher nominal language-model size; HF total includes stored multimodal/MTP tensors as applicable." }, "arch_label": "Dense · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/Qwen/Qwen3.5-27B/blob/fc05daec18b0a78c049392ed2e771dde82bdf654/config.json", "extended": 1010000 }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/Qwen/Qwen3.5-27B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 65536, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3.5-27B/blob/fc05daec18b0a78c049392ed2e771dde82bdf654/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 16740812704, "repo": "unsloth/Qwen3.5-27B-GGUF", "revision": "3221f178a6b842d04f1fb42f1c413534adcc0a6a", "url": "https://huggingface.co/unsloth/Qwen3.5-27B-GGUF/tree/3221f178a6b842d04f1fb42f1c413534adcc0a6a", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 931145984, "url": "https://huggingface.co/unsloth/Qwen3.5-27B-GGUF/blob/3221f178a6b842d04f1fb42f1c413534adcc0a6a/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 927607040, "url": "https://huggingface.co/unsloth/Qwen3.5-27B-GGUF/blob/3221f178a6b842d04f1fb42f1c413534adcc0a6a/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1842940160, "url": "https://huggingface.co/unsloth/Qwen3.5-27B-GGUF/blob/3221f178a6b842d04f1fb42f1c413534adcc0a6a/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 19608995744, "repo": "unsloth/Qwen3.5-27B-GGUF", "revision": "3221f178a6b842d04f1fb42f1c413534adcc0a6a", "url": "https://huggingface.co/unsloth/Qwen3.5-27B-GGUF/tree/3221f178a6b842d04f1fb42f1c413534adcc0a6a", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 931145984, "url": "https://huggingface.co/unsloth/Qwen3.5-27B-GGUF/blob/3221f178a6b842d04f1fb42f1c413534adcc0a6a/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 927607040, "url": "https://huggingface.co/unsloth/Qwen3.5-27B-GGUF/blob/3221f178a6b842d04f1fb42f1c413534adcc0a6a/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1842940160, "url": "https://huggingface.co/unsloth/Qwen3.5-27B-GGUF/blob/3221f178a6b842d04f1fb42f1c413534adcc0a6a/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 28595763104, "repo": "unsloth/Qwen3.5-27B-GGUF", "revision": "3221f178a6b842d04f1fb42f1c413534adcc0a6a", "url": "https://huggingface.co/unsloth/Qwen3.5-27B-GGUF/tree/3221f178a6b842d04f1fb42f1c413534adcc0a6a", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 931145984, "url": "https://huggingface.co/unsloth/Qwen3.5-27B-GGUF/blob/3221f178a6b842d04f1fb42f1c413534adcc0a6a/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 927607040, "url": "https://huggingface.co/unsloth/Qwen3.5-27B-GGUF/blob/3221f178a6b842d04f1fb42f1c413534adcc0a6a/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1842940160, "url": "https://huggingface.co/unsloth/Qwen3.5-27B-GGUF/blob/3221f178a6b842d04f1fb42f1c413534adcc0a6a/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 16054546159, "repo": "mlx-community/Qwen3.5-27B-4bit", "revision": "45797d2985a12c55e6473686e9ea91b95e959553", "url": "https://huggingface.co/mlx-community/Qwen3.5-27B-4bit/tree/45797d2985a12c55e6473686e9ea91b95e959553", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 3, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 29501222821, "repo": "mlx-community/Qwen3.5-27B-8bit", "revision": "d49462a9487464de69e2240d2bd65e50bb7d1cbc", "url": "https://huggingface.co/mlx-community/Qwen3.5-27B-8bit/tree/d49462a9487464de69e2240d2bd65e50bb7d1cbc", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 6, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-03-13", "title": "Chinese-English Typography Issues", "url": "https://huggingface.co/Qwen/Qwen3.5-27B/discussions/32" }, { "platform": "Hugging Face", "date": "2026-03-09", "title": "[transformers-server] Continuous batching does not support Qwen-3.5 series", "url": "https://huggingface.co/Qwen/Qwen3.5-27B/discussions/30" } ], "notes": [] }, { "slug": "qwen3.5-2b", "name": "Qwen3.5-2B", "hf_repo": "Qwen/Qwen3.5-2B", "hf_revision": "15852e8c16360a2fea060d615a32b45270f8a8fc", "source_url": "https://huggingface.co/Qwen/Qwen3.5-2B", "org": "Qwen", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-03-02", "precision": "day", "url": "https://github.com/QwenLM/Qwen3.8/blob/main/README.md", "status": "verified_primary", "note": "Official dated release-news entry states public model availability." }, "params": { "total_b": 2, "active_b": 2, "moe": false, "url": "https://huggingface.co/Qwen/Qwen3.5-2B/blob/15852e8c16360a2fea060d615a32b45270f8a8fc/README.md", "active_url": null, "scope": "Publisher nominal language-model size; HF total includes stored multimodal/MTP tensors as applicable." }, "arch_label": "Dense · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/Qwen/Qwen3.5-2B/blob/15852e8c16360a2fea060d615a32b45270f8a8fc/config.json", "extended": 1010000 }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/Qwen/Qwen3.5-2B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 12288, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3.5-2B/blob/15852e8c16360a2fea060d615a32b45270f8a8fc/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 1280835840, "repo": "unsloth/Qwen3.5-2B-GGUF", "revision": "f6d5376be1edb4d416d56da11e5397a961aca8ae", "url": "https://huggingface.co/unsloth/Qwen3.5-2B-GGUF/tree/f6d5376be1edb4d416d56da11e5397a961aca8ae", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 671372992, "url": "https://huggingface.co/unsloth/Qwen3.5-2B-GGUF/blob/f6d5376be1edb4d416d56da11e5397a961aca8ae/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 668227264, "url": "https://huggingface.co/unsloth/Qwen3.5-2B-GGUF/blob/f6d5376be1edb4d416d56da11e5397a961aca8ae/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1325684416, "url": "https://huggingface.co/unsloth/Qwen3.5-2B-GGUF/blob/f6d5376be1edb4d416d56da11e5397a961aca8ae/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 1435238656, "repo": "unsloth/Qwen3.5-2B-GGUF", "revision": "f6d5376be1edb4d416d56da11e5397a961aca8ae", "url": "https://huggingface.co/unsloth/Qwen3.5-2B-GGUF/tree/f6d5376be1edb4d416d56da11e5397a961aca8ae", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 671372992, "url": "https://huggingface.co/unsloth/Qwen3.5-2B-GGUF/blob/f6d5376be1edb4d416d56da11e5397a961aca8ae/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 668227264, "url": "https://huggingface.co/unsloth/Qwen3.5-2B-GGUF/blob/f6d5376be1edb4d416d56da11e5397a961aca8ae/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1325684416, "url": "https://huggingface.co/unsloth/Qwen3.5-2B-GGUF/blob/f6d5376be1edb4d416d56da11e5397a961aca8ae/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 2012012800, "repo": "unsloth/Qwen3.5-2B-GGUF", "revision": "f6d5376be1edb4d416d56da11e5397a961aca8ae", "url": "https://huggingface.co/unsloth/Qwen3.5-2B-GGUF/tree/f6d5376be1edb4d416d56da11e5397a961aca8ae", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 671372992, "url": "https://huggingface.co/unsloth/Qwen3.5-2B-GGUF/blob/f6d5376be1edb4d416d56da11e5397a961aca8ae/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 668227264, "url": "https://huggingface.co/unsloth/Qwen3.5-2B-GGUF/blob/f6d5376be1edb4d416d56da11e5397a961aca8ae/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1325684416, "url": "https://huggingface.co/unsloth/Qwen3.5-2B-GGUF/blob/f6d5376be1edb4d416d56da11e5397a961aca8ae/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_FP16": { "bytes": 3775709120, "repo": "AaryanK/Qwen3.5-2B-GGUF", "revision": "3b3b8686d74f2ccfff1bcf36f80e114aff1b7881", "url": "https://huggingface.co/AaryanK/Qwen3.5-2B-GGUF/tree/3b3b8686d74f2ccfff1bcf36f80e114aff1b7881", "precision": "FP16", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 1722271785, "repo": "mlx-community/Qwen3.5-2B-4bit", "revision": "674aaa7240b91e8012fcad5d791b7dfe5ba90207", "url": "https://huggingface.co/mlx-community/Qwen3.5-2B-4bit/tree/674aaa7240b91e8012fcad5d791b7dfe5ba90207", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 2662910403, "repo": "mlx-community/Qwen3.5-2B-8bit", "revision": "562fb8ea19bbe7565a3ddf5b7d49899ea2c88d2b", "url": "https://huggingface.co/mlx-community/Qwen3.5-2B-8bit/tree/562fb8ea19bbe7565a3ddf5b7d49899ea2c88d2b", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 1, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [], "community": [ { "platform": "Hugging Face", "date": "2026-08-30", "title": "Run Qwen 3.5 in C under 8 GB RAM", "url": "https://huggingface.co/Qwen/Qwen3.5-2B/discussions/16" }, { "platform": "Hugging Face", "date": "2026-08-08", "title": "Listed on OpenModelMap", "url": "https://huggingface.co/Qwen/Qwen3.5-2B/discussions/12" } ], "notes": [] }, { "slug": "qwen3.5-35b-a3b", "name": "Qwen3.5-35B-A3B", "hf_repo": "Qwen/Qwen3.5-35B-A3B", "hf_revision": "59d61f3ce65a6d9863b86d2e96597125219dc754", "source_url": "https://huggingface.co/Qwen/Qwen3.5-35B-A3B", "org": "Qwen", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-02-24", "precision": "day", "url": "https://github.com/QwenLM/Qwen3.8/blob/main/README.md", "status": "verified_primary", "note": "Official dated release-news entry states public model availability." }, "params": { "total_b": 35, "active_b": 3, "moe": true, "url": "https://huggingface.co/Qwen/Qwen3.5-35B-A3B/blob/59d61f3ce65a6d9863b86d2e96597125219dc754/README.md", "active_url": "https://huggingface.co/Qwen/Qwen3.5-35B-A3B/blob/59d61f3ce65a6d9863b86d2e96597125219dc754/README.md", "scope": "Publisher nominal language-model size; HF total includes stored multimodal/MTP tensors as applicable." }, "arch_label": "MoE · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/Qwen/Qwen3.5-35B-A3B/blob/59d61f3ce65a6d9863b86d2e96597125219dc754/config.json", "extended": 1010000 }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/Qwen/Qwen3.5-35B-A3B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 20480, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3.5-35B-A3B/blob/59d61f3ce65a6d9863b86d2e96597125219dc754/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 22016023168, "repo": "unsloth/Qwen3.5-35B-A3B-GGUF", "revision": "bc014a17be43adabd7066b7a86075ff935c6a4e2", "url": "https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF/tree/bc014a17be43adabd7066b7a86075ff935c6a4e2", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 902822592, "url": "https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF/blob/bc014a17be43adabd7066b7a86075ff935c6a4e2/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 899283648, "url": "https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF/blob/bc014a17be43adabd7066b7a86075ff935c6a4e2/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1786305216, "url": "https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF/blob/bc014a17be43adabd7066b7a86075ff935c6a4e2/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 26249607808, "repo": "unsloth/Qwen3.5-35B-A3B-GGUF", "revision": "bc014a17be43adabd7066b7a86075ff935c6a4e2", "url": "https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF/tree/bc014a17be43adabd7066b7a86075ff935c6a4e2", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 902822592, "url": "https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF/blob/bc014a17be43adabd7066b7a86075ff935c6a4e2/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 899283648, "url": "https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF/blob/bc014a17be43adabd7066b7a86075ff935c6a4e2/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1786305216, "url": "https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF/blob/bc014a17be43adabd7066b7a86075ff935c6a4e2/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 36903139968, "repo": "unsloth/Qwen3.5-35B-A3B-GGUF", "revision": "bc014a17be43adabd7066b7a86075ff935c6a4e2", "url": "https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF/tree/bc014a17be43adabd7066b7a86075ff935c6a4e2", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 902822592, "url": "https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF/blob/bc014a17be43adabd7066b7a86075ff935c6a4e2/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 899283648, "url": "https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF/blob/bc014a17be43adabd7066b7a86075ff935c6a4e2/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1786305216, "url": "https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF/blob/bc014a17be43adabd7066b7a86075ff935c6a4e2/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 20391679439, "repo": "mlx-community/Qwen3.5-35B-A3B-4bit", "revision": "1e20fd8d42056f870933bf98ca6211024744f7ec", "url": "https://huggingface.co/mlx-community/Qwen3.5-35B-A3B-4bit/tree/1e20fd8d42056f870933bf98ca6211024744f7ec", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 4, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 37721404892, "repo": "mlx-community/Qwen3.5-35B-A3B-8bit", "revision": "690e046ad0a71d206b054bb63b154e5106a91132", "url": "https://huggingface.co/mlx-community/Qwen3.5-35B-A3B-8bit/tree/690e046ad0a71d206b054bb63b154e5106a91132", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 8, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-06-02", "title": "Given a search question, return reasoning and content(\"\\n\\n\") with no tool calls in first turn ending with finish_reason==stop.", "url": "https://huggingface.co/Qwen/Qwen3.5-35B-A3B/discussions/77" }, { "platform": "Hugging Face", "date": "2026-04-12", "title": "Compact tool-rendering template — 49% fewer prompt tokens", "url": "https://huggingface.co/Qwen/Qwen3.5-35B-A3B/discussions/72" } ], "notes": [] }, { "slug": "qwen3.6-27b", "name": "Qwen3.6-27B", "hf_repo": "Qwen/Qwen3.6-27B", "hf_revision": "6a9e13bd6fc8f0983b9b99948120bc37f49c13e9", "source_url": "https://huggingface.co/Qwen/Qwen3.6-27B", "org": "Qwen", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-04-22", "precision": "day", "url": "https://github.com/QwenLM/Qwen3.8/blob/main/README.md", "status": "verified_primary", "note": "Official dated release-news entry states public model availability." }, "params": { "total_b": 27, "active_b": 27, "moe": false, "url": "https://huggingface.co/Qwen/Qwen3.6-27B/blob/6a9e13bd6fc8f0983b9b99948120bc37f49c13e9/README.md", "active_url": null, "scope": "Publisher nominal language-model size; HF total includes stored multimodal/MTP tensors as applicable." }, "arch_label": "Dense · hybrid linear attn", "context": { "native": 262144, "url": "https://huggingface.co/Qwen/Qwen3.6-27B/blob/6a9e13bd6fc8f0983b9b99948120bc37f49c13e9/config.json", "extended": 1010000 }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/Qwen/Qwen3.6-27B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 65536, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3.6-27B/blob/6a9e13bd6fc8f0983b9b99948120bc37f49c13e9/config.json", "dtype": "FP16", "note": "Only full-attention KV grows with sequence length; hybrid recurrent/convolution state adds memory. Sliding cache saturation requires backend support." }, "artifacts": { "GGUF_Q4_K_M": { "bytes": 16817244384, "repo": "unsloth/Qwen3.6-27B-GGUF", "revision": "82d411acf4a06cfb8d9b073a5211bf410bfc29bf", "url": "https://huggingface.co/unsloth/Qwen3.6-27B-GGUF/tree/82d411acf4a06cfb8d9b073a5211bf410bfc29bf", "precision": "Q4_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 931146304, "url": "https://huggingface.co/unsloth/Qwen3.6-27B-GGUF/blob/82d411acf4a06cfb8d9b073a5211bf410bfc29bf/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 927607360, "url": "https://huggingface.co/unsloth/Qwen3.6-27B-GGUF/blob/82d411acf4a06cfb8d9b073a5211bf410bfc29bf/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1842940480, "url": "https://huggingface.co/unsloth/Qwen3.6-27B-GGUF/blob/82d411acf4a06cfb8d9b073a5211bf410bfc29bf/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q5_K_M": { "bytes": 19509790944, "repo": "unsloth/Qwen3.6-27B-GGUF", "revision": "82d411acf4a06cfb8d9b073a5211bf410bfc29bf", "url": "https://huggingface.co/unsloth/Qwen3.6-27B-GGUF/tree/82d411acf4a06cfb8d9b073a5211bf410bfc29bf", "precision": "Q5_K_M", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 931146304, "url": "https://huggingface.co/unsloth/Qwen3.6-27B-GGUF/blob/82d411acf4a06cfb8d9b073a5211bf410bfc29bf/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 927607360, "url": "https://huggingface.co/unsloth/Qwen3.6-27B-GGUF/blob/82d411acf4a06cfb8d9b073a5211bf410bfc29bf/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1842940480, "url": "https://huggingface.co/unsloth/Qwen3.6-27B-GGUF/blob/82d411acf4a06cfb8d9b073a5211bf410bfc29bf/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "GGUF_Q8_0": { "bytes": 28595763424, "repo": "unsloth/Qwen3.6-27B-GGUF", "revision": "82d411acf4a06cfb8d9b073a5211bf410bfc29bf", "url": "https://huggingface.co/unsloth/Qwen3.6-27B-GGUF/tree/82d411acf4a06cfb8d9b073a5211bf410bfc29bf", "precision": "Q8_0", "scope": "text/backbone GGUF; separate projectors listed below", "files": 1, "status": "verified_hf_file_sizes", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 931146304, "url": "https://huggingface.co/unsloth/Qwen3.6-27B-GGUF/blob/82d411acf4a06cfb8d9b073a5211bf410bfc29bf/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 927607360, "url": "https://huggingface.co/unsloth/Qwen3.6-27B-GGUF/blob/82d411acf4a06cfb8d9b073a5211bf410bfc29bf/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 1842940480, "url": "https://huggingface.co/unsloth/Qwen3.6-27B-GGUF/blob/82d411acf4a06cfb8d9b073a5211bf410bfc29bf/mmproj-F32.gguf" } ], "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_4bit": { "bytes": 16054541599, "repo": "mlx-community/Qwen3.6-27B-4bit", "revision": "c000ac2c2057d94be3fa931000c31723aac53282", "url": "https://huggingface.co/mlx-community/Qwen3.6-27B-4bit/tree/c000ac2c2057d94be3fa931000c31723aac53282", "precision": "4-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 3, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." }, "MLX_8bit": { "bytes": 29501218261, "repo": "mlx-community/Qwen3.6-27B-8bit", "revision": "c5a593c1475a746e43a543b0a02bd2b357e5745f", "url": "https://huggingface.co/mlx-community/Qwen3.6-27B-8bit/tree/c5a593c1475a746e43a543b0a02bd2b357e5745f", "precision": "8-bit MLX", "scope": "whole MLX safetensors checkpoint; included modalities follow conversion", "files": 6, "status": "verified_hf_file_sizes", "note": "Artifact bytes, not allocated VRAM; includes serialization metadata. Weight precision can be mixed." } }, "artifacts_missing": [ "GGUF_FP16" ], "community": [ { "platform": "Hugging Face", "date": "2026-05-01", "title": "GGUF available — Cerebellum v4 (ablation-guided mixed-precision)", "url": "https://huggingface.co/Qwen/Qwen3.6-27B/discussions/28" }, { "platform": "Hugging Face", "date": "2026-04-24", "title": "Qwen3.6 27b has template issue in ralph loop on opencode: `No user query found in messages`", "url": "https://huggingface.co/Qwen/Qwen3.6-27B/discussions/16" } ], "notes": [] }, { "slug": "step-3.7-flash", "name": "Step 3.7 Flash", "hf_repo": "stepfun-ai/Step-3.7-Flash", "hf_revision": "5f6244077ac62e04eec3f320501ff8c2b293373a", "source_url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash", "org": "stepfun-ai", "status": "current", "catalog_status": "current_2026", "release_2026_verified": true, "default_visible": true, "release": { "label": "2026-05-29", "precision": "day", "url": "https://static.stepfun.com/blog/step-3.7-flash/", "status": "verified_primary", "note": "Official release blog links released Hugging Face model." }, "params": { "total_b": 198, "active_b": 11, "moe": true, "url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash/raw/5f6244077ac62e04eec3f320501ff8c2b293373a/README.md", "active_url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash/raw/5f6244077ac62e04eec3f320501ff8c2b293373a/README.md", "scope": null }, "arch_label": "MoE", "context": { "native": 262144, "url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash/raw/5f6244077ac62e04eec3f320501ff8c2b293373a/config.json", "extended": null }, "license": { "value": "apache-2.0", "url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash/blob/5f6244077ac62e04eec3f320501ff8c2b293373a/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": null, "win_per_token": null, "win_tokens": null, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "unknown", "url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash/raw/5f6244077ac62e04eec3f320501ff8c2b293373a/config.json", "dtype": "BF16", "note": "Architecture needs backend-specific cache/state accounting; do not apply plain GQA formula." }, "artifacts": { "GGUF_Q8_0": { "bytes": 209417872192, "repo": "unsloth/Step-3.7-Flash-GGUF", "revision": "8dd62fe8a98f3c8994160d0c521542560b558f37", "url": "https://huggingface.co/unsloth/Step-3.7-Flash-GGUF/tree/8dd62fe8a98f3c8994160d0c521542560b558f37", "precision": "GGUF Q8_0", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 6, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 4399309024, "url": "https://huggingface.co/unsloth/Step-3.7-Flash-GGUF/resolve/8dd62fe8a98f3c8994160d0c521542560b558f37/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 3972829408, "url": "https://huggingface.co/unsloth/Step-3.7-Flash-GGUF/resolve/8dd62fe8a98f3c8994160d0c521542560b558f37/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 7924097248, "url": "https://huggingface.co/unsloth/Step-3.7-Flash-GGUF/resolve/8dd62fe8a98f3c8994160d0c521542560b558f37/mmproj-F32.gguf" } ] }, "GGUF_BF16": { "bytes": 394017424160, "repo": "unsloth/Step-3.7-Flash-GGUF", "revision": "8dd62fe8a98f3c8994160d0c521542560b558f37", "url": "https://huggingface.co/unsloth/Step-3.7-Flash-GGUF/tree/8dd62fe8a98f3c8994160d0c521542560b558f37", "precision": "GGUF BF16", "scope": "GGUF model weights; separate mmproj/projector files excluded", "files": 8, "status": "measured_file_manifest", "projectors": [ { "path": "mmproj-BF16.gguf", "bytes": 4399309024, "url": "https://huggingface.co/unsloth/Step-3.7-Flash-GGUF/resolve/8dd62fe8a98f3c8994160d0c521542560b558f37/mmproj-BF16.gguf" }, { "path": "mmproj-F16.gguf", "bytes": 3972829408, "url": "https://huggingface.co/unsloth/Step-3.7-Flash-GGUF/resolve/8dd62fe8a98f3c8994160d0c521542560b558f37/mmproj-F16.gguf" }, { "path": "mmproj-F32.gguf", "bytes": 7924097248, "url": "https://huggingface.co/unsloth/Step-3.7-Flash-GGUF/resolve/8dd62fe8a98f3c8994160d0c521542560b558f37/mmproj-F32.gguf" } ] }, "MLX_4bit": { "bytes": 110813433037, "repo": "mlx-community/Step-3.7-Flash-4bit", "revision": "123635da9dc9e06250f5150e8773a5a1992edc58", "url": "https://huggingface.co/mlx-community/Step-3.7-Flash-4bit/tree/123635da9dc9e06250f5150e8773a5a1992edc58", "precision": "MLX nominal 4-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 22, "status": "measured_file_manifest", "mixed": true }, "MLX_8bit": { "bytes": 209266529037, "repo": "mlx-community/Step-3.7-Flash-8bit", "revision": "a4efa6b0152babde095a10208a7ddc7f44d91346", "url": "https://huggingface.co/mlx-community/Step-3.7-Flash-8bit/tree/a4efa6b0152babde095a10208a7ddc7f44d91346", "precision": "MLX nominal 8-bit", "scope": "Converted MLX safetensors model shard set; may preserve some tensors at higher precision", "files": 43, "status": "measured_file_manifest", "mixed": true } }, "artifacts_missing": [ "GGUF_Q4_K_M", "GGUF_Q5_K_M", "GGUF_FP16" ], "community": [], "notes": [ "196B language backbone plus 1.8B vision encoder, publisher rounded total." ] }, { "slug": "qwen3-0.6b", "name": "Qwen3-0.6B", "hf_repo": "Qwen/Qwen3-0.6B", "hf_revision": null, "source_url": "https://huggingface.co/Qwen/Qwen3-0.6B", "org": "Qwen", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-04-29", "precision": "day", "url": "https://qwenlm.github.io/blog/qwen3/", "status": "verified_primary", "note": "Original Qwen3 release; later Coder and Next variants have separate dates." }, "params": { "total_b": 0.6, "active_b": 0.6, "moe": false, "url": "https://huggingface.co/Qwen/Qwen3-0.6B/raw/main/README.md", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 32768, "url": "https://huggingface.co/Qwen/Qwen3-0.6B/raw/main/README.md", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/Qwen/Qwen3-0.6B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 114688, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3-0.6B/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "qwen3-1.7b", "name": "Qwen3-1.7B", "hf_repo": "Qwen/Qwen3-1.7B", "hf_revision": null, "source_url": "https://huggingface.co/Qwen/Qwen3-1.7B", "org": "Qwen", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-04-29", "precision": "day", "url": "https://qwenlm.github.io/blog/qwen3/", "status": "verified_primary", "note": "Original Qwen3 release; later Coder and Next variants have separate dates." }, "params": { "total_b": 1.7, "active_b": 1.7, "moe": false, "url": "https://huggingface.co/Qwen/Qwen3-1.7B/raw/main/README.md", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 32768, "url": "https://huggingface.co/Qwen/Qwen3-1.7B/raw/main/README.md", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/Qwen/Qwen3-1.7B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 114688, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3-1.7B/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "qwen3-4b", "name": "Qwen3-4B", "hf_repo": "Qwen/Qwen3-4B", "hf_revision": null, "source_url": "https://huggingface.co/Qwen/Qwen3-4B", "org": "Qwen", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-04-29", "precision": "day", "url": "https://qwenlm.github.io/blog/qwen3/", "status": "verified_primary", "note": "Original Qwen3 release; later Coder and Next variants have separate dates." }, "params": { "total_b": 4.022, "active_b": 4.022, "moe": false, "url": "https://huggingface.co/api/models/Qwen/Qwen3-4B", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 32768, "url": "https://huggingface.co/Qwen/Qwen3-4B/raw/main/README.md", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/Qwen/Qwen3-4B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 147456, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3-4B/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "qwen3-8b", "name": "Qwen3-8B", "hf_repo": "Qwen/Qwen3-8B", "hf_revision": null, "source_url": "https://huggingface.co/Qwen/Qwen3-8B", "org": "Qwen", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-04-29", "precision": "day", "url": "https://qwenlm.github.io/blog/qwen3/", "status": "verified_primary", "note": "Original Qwen3 release; later Coder and Next variants have separate dates." }, "params": { "total_b": 8.191, "active_b": 8.191, "moe": false, "url": "https://huggingface.co/api/models/Qwen/Qwen3-8B", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 32768, "url": "https://huggingface.co/Qwen/Qwen3-8B/raw/main/README.md", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/Qwen/Qwen3-8B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 147456, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3-8B/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "qwen3-14b", "name": "Qwen3-14B", "hf_repo": "Qwen/Qwen3-14B", "hf_revision": null, "source_url": "https://huggingface.co/Qwen/Qwen3-14B", "org": "Qwen", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-04-29", "precision": "day", "url": "https://qwenlm.github.io/blog/qwen3/", "status": "verified_primary", "note": "Original Qwen3 release; later Coder and Next variants have separate dates." }, "params": { "total_b": 14.768, "active_b": 14.768, "moe": false, "url": "https://huggingface.co/api/models/Qwen/Qwen3-14B", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 32768, "url": "https://huggingface.co/Qwen/Qwen3-14B/raw/main/README.md", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/Qwen/Qwen3-14B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 163840, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3-14B/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "qwen3-32b", "name": "Qwen3-32B", "hf_repo": "Qwen/Qwen3-32B", "hf_revision": null, "source_url": "https://huggingface.co/Qwen/Qwen3-32B", "org": "Qwen", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-04-29", "precision": "day", "url": "https://qwenlm.github.io/blog/qwen3/", "status": "verified_primary", "note": "Original Qwen3 release; later Coder and Next variants have separate dates." }, "params": { "total_b": 32.762, "active_b": 32.762, "moe": false, "url": "https://huggingface.co/api/models/Qwen/Qwen3-32B", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 32768, "url": "https://huggingface.co/Qwen/Qwen3-32B/raw/main/README.md", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/Qwen/Qwen3-32B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 262144, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3-32B/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "qwen3-30b-a3b", "name": "Qwen3-30B-A3B", "hf_repo": "Qwen/Qwen3-30B-A3B", "hf_revision": null, "source_url": "https://huggingface.co/Qwen/Qwen3-30B-A3B", "org": "Qwen", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-04-29", "precision": "day", "url": "https://qwenlm.github.io/blog/qwen3/", "status": "verified_primary", "note": "Original Qwen3 release; later Coder and Next variants have separate dates." }, "params": { "total_b": 30.532, "active_b": 3.3, "moe": true, "url": "https://huggingface.co/api/models/Qwen/Qwen3-30B-A3B", "active_url": "https://huggingface.co/Qwen/Qwen3-30B-A3B/raw/main/README.md", "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "MoE", "context": { "native": 32768, "url": "https://huggingface.co/Qwen/Qwen3-30B-A3B/raw/main/README.md", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/Qwen/Qwen3-30B-A3B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 98304, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3-30B-A3B/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "qwen3-235b-a22b", "name": "Qwen3-235B-A22B", "hf_repo": "Qwen/Qwen3-235B-A22B", "hf_revision": null, "source_url": "https://huggingface.co/Qwen/Qwen3-235B-A22B", "org": "Qwen", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-04-29", "precision": "day", "url": "https://qwenlm.github.io/blog/qwen3/", "status": "verified_primary", "note": "Original Qwen3 release; later Coder and Next variants have separate dates." }, "params": { "total_b": 235.094, "active_b": 22, "moe": true, "url": "https://huggingface.co/api/models/Qwen/Qwen3-235B-A22B", "active_url": "https://huggingface.co/Qwen/Qwen3-235B-A22B/raw/main/README.md", "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "MoE", "context": { "native": 32768, "url": "https://huggingface.co/Qwen/Qwen3-235B-A22B/raw/main/README.md", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/Qwen/Qwen3-235B-A22B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 192512, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3-235B-A22B/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "qwen3-coder-30b-a3b", "name": "Qwen3-Coder-30B-A3B-Instruct", "hf_repo": "Qwen/Qwen3-Coder-30B-A3B-Instruct", "hf_revision": null, "source_url": "https://huggingface.co/Qwen/Qwen3-Coder-30B-A3B-Instruct", "org": "Qwen", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-07-31", "precision": "day", "url": "https://docs.aws.amazon.com/bedrock/latest/userguide/model-card-qwen-qwen3-coder-30b-a3b-instruct.html", "status": "verified_primary", "note": "Primary deployment-provider model card launch date; publisher dated announcement not recovered." }, "params": { "total_b": 30.532, "active_b": 3.3, "moe": true, "url": "https://huggingface.co/api/models/Qwen/Qwen3-Coder-30B-A3B-Instruct", "active_url": "https://huggingface.co/Qwen/Qwen3-Coder-30B-A3B-Instruct/raw/main/README.md", "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "MoE", "context": { "native": 262144, "url": "https://huggingface.co/Qwen/Qwen3-Coder-30B-A3B-Instruct/raw/main/README.md", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/Qwen/Qwen3-Coder-30B-A3B-Instruct/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 98304, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3-Coder-30B-A3B-Instruct/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "qwen3-coder-480b-a35b", "name": "Qwen3-Coder-480B-A35B-Instruct", "hf_repo": "Qwen/Qwen3-Coder-480B-A35B-Instruct", "hf_revision": null, "source_url": "https://huggingface.co/Qwen/Qwen3-Coder-480B-A35B-Instruct", "org": "Qwen", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-07-22", "precision": "day", "url": "https://qwenlm.github.io/blog/qwen3-coder/", "status": "verified_primary", "note": null }, "params": { "total_b": 480.155, "active_b": 35, "moe": true, "url": "https://huggingface.co/api/models/Qwen/Qwen3-Coder-480B-A35B-Instruct", "active_url": "https://huggingface.co/Qwen/Qwen3-Coder-480B-A35B-Instruct/raw/main/README.md", "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "MoE", "context": { "native": 262144, "url": "https://huggingface.co/Qwen/Qwen3-Coder-480B-A35B-Instruct/raw/main/README.md", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/Qwen/Qwen3-Coder-480B-A35B-Instruct/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 253952, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3-Coder-480B-A35B-Instruct/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "qwen3-next-80b-a3b", "name": "Qwen3-Next-80B-A3B-Instruct", "hf_repo": "Qwen/Qwen3-Next-80B-A3B-Instruct", "hf_revision": null, "source_url": "https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Instruct", "org": "Qwen", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-09-11", "precision": "day", "url": "https://raw.githubusercontent.com/QwenLM/Qwen3-Next/main/README.md", "status": "verified_primary", "note": null }, "params": { "total_b": 81.325, "active_b": 3, "moe": true, "url": "https://huggingface.co/api/models/Qwen/Qwen3-Next-80B-A3B-Instruct", "active_url": "https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Instruct/raw/main/README.md", "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "MoE", "context": { "native": 262144, "url": "https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Instruct/raw/main/README.md", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Instruct/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 24576, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": null, "unknown": [ "fixed_state" ], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Instruct/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "qwen2.5-7b", "name": "Qwen2.5-7B-Instruct", "hf_repo": "Qwen/Qwen2.5-7B-Instruct", "hf_revision": null, "source_url": "https://huggingface.co/Qwen/Qwen2.5-7B-Instruct", "org": "Qwen", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2024-09-19", "precision": "day", "url": "https://qwenlm.github.io/blog/qwen2.5/", "status": "verified_primary", "note": null }, "params": { "total_b": 7.616, "active_b": 7.616, "moe": false, "url": "https://huggingface.co/api/models/Qwen/Qwen2.5-7B-Instruct", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 32768, "url": "https://huggingface.co/Qwen/Qwen2.5-7B-Instruct/resolve/main/config.json", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/Qwen/Qwen2.5-7B-Instruct/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 57344, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen2.5-7B-Instruct/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "qwen2.5-14b", "name": "Qwen2.5-14B-Instruct", "hf_repo": "Qwen/Qwen2.5-14B-Instruct", "hf_revision": null, "source_url": "https://huggingface.co/Qwen/Qwen2.5-14B-Instruct", "org": "Qwen", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2024-09-19", "precision": "day", "url": "https://qwenlm.github.io/blog/qwen2.5/", "status": "verified_primary", "note": null }, "params": { "total_b": 14.77, "active_b": 14.77, "moe": false, "url": "https://huggingface.co/api/models/Qwen/Qwen2.5-14B-Instruct", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 32768, "url": "https://huggingface.co/Qwen/Qwen2.5-14B-Instruct/resolve/main/config.json", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/Qwen/Qwen2.5-14B-Instruct/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 196608, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen2.5-14B-Instruct/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "qwen2.5-32b", "name": "Qwen2.5-32B-Instruct", "hf_repo": "Qwen/Qwen2.5-32B-Instruct", "hf_revision": null, "source_url": "https://huggingface.co/Qwen/Qwen2.5-32B-Instruct", "org": "Qwen", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2024-09-19", "precision": "day", "url": "https://qwenlm.github.io/blog/qwen2.5/", "status": "verified_primary", "note": null }, "params": { "total_b": 32.764, "active_b": 32.764, "moe": false, "url": "https://huggingface.co/api/models/Qwen/Qwen2.5-32B-Instruct", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 32768, "url": "https://huggingface.co/Qwen/Qwen2.5-32B-Instruct/resolve/main/config.json", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/Qwen/Qwen2.5-32B-Instruct/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 262144, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen2.5-32B-Instruct/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "qwen2.5-72b", "name": "Qwen2.5-72B-Instruct", "hf_repo": "Qwen/Qwen2.5-72B-Instruct", "hf_revision": null, "source_url": "https://huggingface.co/Qwen/Qwen2.5-72B-Instruct", "org": "Qwen", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2024-09-19", "precision": "day", "url": "https://qwenlm.github.io/blog/qwen2.5/", "status": "verified_primary", "note": null }, "params": { "total_b": 72.706, "active_b": 72.706, "moe": false, "url": "https://huggingface.co/api/models/Qwen/Qwen2.5-72B-Instruct", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 32768, "url": "https://huggingface.co/Qwen/Qwen2.5-72B-Instruct/resolve/main/config.json", "extended": null }, "license": { "value": "Qwen License", "url": "https://huggingface.co/Qwen/Qwen2.5-72B-Instruct/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 327680, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen2.5-72B-Instruct/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "qwen2.5-coder-7b", "name": "Qwen2.5-Coder-7B-Instruct", "hf_repo": "Qwen/Qwen2.5-Coder-7B-Instruct", "hf_revision": null, "source_url": "https://huggingface.co/Qwen/Qwen2.5-Coder-7B-Instruct", "org": "Qwen", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2024-09-19", "precision": "day", "url": "https://qwenlm.github.io/blog/qwen2.5-coder/", "status": "verified_primary", "note": null }, "params": { "total_b": 7.616, "active_b": 7.616, "moe": false, "url": "https://huggingface.co/api/models/Qwen/Qwen2.5-Coder-7B-Instruct", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 32768, "url": "https://huggingface.co/Qwen/Qwen2.5-Coder-7B-Instruct/resolve/main/config.json", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/Qwen/Qwen2.5-Coder-7B-Instruct/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 57344, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen2.5-Coder-7B-Instruct/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "qwen2.5-coder-32b", "name": "Qwen2.5-Coder-32B-Instruct", "hf_repo": "Qwen/Qwen2.5-Coder-32B-Instruct", "hf_revision": null, "source_url": "https://huggingface.co/Qwen/Qwen2.5-Coder-32B-Instruct", "org": "Qwen", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2024-11-12", "precision": "day", "url": "https://qwenlm.github.io/blog/qwen2.5-coder-family/", "status": "verified_primary", "note": "Date shown by publisher; timezone-local announcement day can differ from UTC weights upload." }, "params": { "total_b": 32.764, "active_b": 32.764, "moe": false, "url": "https://huggingface.co/api/models/Qwen/Qwen2.5-Coder-32B-Instruct", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 32768, "url": "https://huggingface.co/Qwen/Qwen2.5-Coder-32B-Instruct/resolve/main/config.json", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/Qwen/Qwen2.5-Coder-32B-Instruct/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 262144, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/Qwen/Qwen2.5-Coder-32B-Instruct/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "llama-2-7b", "name": "Llama 2 7B", "hf_repo": "meta-llama/Llama-2-7b-hf", "hf_revision": null, "source_url": "https://huggingface.co/meta-llama/Llama-2-7b-hf", "org": "meta-llama", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2023-07-18", "precision": "day", "url": "https://raw.githubusercontent.com/meta-llama/llama-models/main/README.md", "status": "verified_primary", "note": null }, "params": { "total_b": 6.738, "active_b": 6.738, "moe": false, "url": "https://huggingface.co/api/models/meta-llama/Llama-2-7b-hf", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 4096, "url": "https://huggingface.co/NousResearch/Llama-2-7b-hf/resolve/main/config.json", "extended": null }, "license": { "value": "Llama 2 Community License", "url": "https://huggingface.co/meta-llama/Llama-2-7b-hf/blob/main/LICENSE.txt" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 524288, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/NousResearch/Llama-2-7b-hf/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "llama-3.1-8b", "name": "Llama 3.1 8B Instruct", "hf_repo": "meta-llama/Llama-3.1-8B-Instruct", "hf_revision": null, "source_url": "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct", "org": "meta-llama", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2024-07-23", "precision": "day", "url": "https://raw.githubusercontent.com/meta-llama/llama-models/main/README.md", "status": "verified_primary", "note": null }, "params": { "total_b": 8.03, "active_b": 8.03, "moe": false, "url": "https://huggingface.co/api/models/meta-llama/Llama-3.1-8B-Instruct", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/unsloth/Llama-3.1-8B-Instruct/resolve/main/config.json", "extended": null }, "license": { "value": "Llama 3.1 Community License", "url": "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 131072, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/unsloth/Llama-3.1-8B-Instruct/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "llama-3.1-70b", "name": "Llama 3.1 70B Instruct", "hf_repo": "meta-llama/Llama-3.1-70B-Instruct", "hf_revision": null, "source_url": "https://huggingface.co/meta-llama/Llama-3.1-70B-Instruct", "org": "meta-llama", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2024-07-23", "precision": "day", "url": "https://raw.githubusercontent.com/meta-llama/llama-models/main/README.md", "status": "verified_primary", "note": null }, "params": { "total_b": 70.554, "active_b": 70.554, "moe": false, "url": "https://huggingface.co/api/models/meta-llama/Llama-3.1-70B-Instruct", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/unsloth/Meta-Llama-3.1-70B-Instruct/resolve/main/config.json", "extended": null }, "license": { "value": "Llama 3.1 Community License", "url": "https://huggingface.co/meta-llama/Llama-3.1-70B-Instruct/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 327680, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/unsloth/Meta-Llama-3.1-70B-Instruct/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "llama-3.2-1b", "name": "Llama 3.2 1B Instruct", "hf_repo": "meta-llama/Llama-3.2-1B-Instruct", "hf_revision": null, "source_url": "https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct", "org": "meta-llama", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2024-09-25", "precision": "day", "url": "https://raw.githubusercontent.com/meta-llama/llama-models/main/README.md", "status": "verified_primary", "note": null }, "params": { "total_b": 1.236, "active_b": 1.236, "moe": false, "url": "https://huggingface.co/api/models/meta-llama/Llama-3.2-1B-Instruct", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/unsloth/Llama-3.2-1B-Instruct/resolve/main/config.json", "extended": null }, "license": { "value": "Llama 3.2 Community License", "url": "https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct/blob/main/LICENSE.txt" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 32768, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/unsloth/Llama-3.2-1B-Instruct/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "llama-3.2-3b", "name": "Llama 3.2 3B Instruct", "hf_repo": "meta-llama/Llama-3.2-3B-Instruct", "hf_revision": null, "source_url": "https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct", "org": "meta-llama", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2024-09-25", "precision": "day", "url": "https://raw.githubusercontent.com/meta-llama/llama-models/main/README.md", "status": "verified_primary", "note": null }, "params": { "total_b": 3.213, "active_b": 3.213, "moe": false, "url": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/unsloth/Llama-3.2-3B-Instruct/resolve/main/config.json", "extended": null }, "license": { "value": "Llama 3.2 Community License", "url": "https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct/blob/main/LICENSE.txt" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 114688, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/unsloth/Llama-3.2-3B-Instruct/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "llama-3.3-70b", "name": "Llama 3.3 70B Instruct", "hf_repo": "meta-llama/Llama-3.3-70B-Instruct", "hf_revision": null, "source_url": "https://huggingface.co/meta-llama/Llama-3.3-70B-Instruct", "org": "meta-llama", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2024-12-06", "precision": "day", "url": "https://raw.githubusercontent.com/meta-llama/llama-models/main/models/llama3_3/MODEL_CARD.md", "status": "verified_primary", "note": "Official model card release date takes precedence over repository overview table, which incorrectly/differently lists December 4." }, "params": { "total_b": 70.554, "active_b": 70.554, "moe": false, "url": "https://huggingface.co/api/models/meta-llama/Llama-3.3-70B-Instruct", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/unsloth/Llama-3.3-70B-Instruct/resolve/main/config.json", "extended": null }, "license": { "value": "Llama 3.3 Community License", "url": "https://huggingface.co/meta-llama/Llama-3.3-70B-Instruct/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 327680, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/unsloth/Llama-3.3-70B-Instruct/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "llama-4-scout", "name": "Llama 4 Scout 17B-16E Instruct", "hf_repo": "meta-llama/Llama-4-Scout-17B-16E-Instruct", "hf_revision": null, "source_url": "https://huggingface.co/meta-llama/Llama-4-Scout-17B-16E-Instruct", "org": "meta-llama", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-04-05", "precision": "day", "url": "https://raw.githubusercontent.com/meta-llama/llama-models/main/README.md", "status": "verified_primary", "note": null }, "params": { "total_b": 108.642, "active_b": 17, "moe": true, "url": "https://huggingface.co/api/models/meta-llama/Llama-4-Scout-17B-16E-Instruct", "active_url": "https://huggingface.co/meta-llama/Llama-4-Scout-17B-16E-Instruct", "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "MoE", "context": { "native": 10485760, "url": "https://huggingface.co/unsloth/Llama-4-Scout-17B-16E-Instruct/resolve/main/config.json", "extended": null }, "license": { "value": "Llama 4 Community License", "url": "https://huggingface.co/meta-llama/Llama-4-Scout-17B-16E-Instruct/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 49152, "win_per_token": 147456, "win_tokens": 8192, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/unsloth/Llama-4-Scout-17B-16E-Instruct/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "llama-4-maverick", "name": "Llama 4 Maverick 17B-128E Instruct", "hf_repo": "meta-llama/Llama-4-Maverick-17B-128E-Instruct", "hf_revision": null, "source_url": "https://huggingface.co/meta-llama/Llama-4-Maverick-17B-128E-Instruct", "org": "meta-llama", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-04-05", "precision": "day", "url": "https://raw.githubusercontent.com/meta-llama/llama-models/main/README.md", "status": "verified_primary", "note": null }, "params": { "total_b": 401.584, "active_b": 17, "moe": true, "url": "https://huggingface.co/api/models/meta-llama/Llama-4-Maverick-17B-128E-Instruct", "active_url": "https://huggingface.co/meta-llama/Llama-4-Maverick-17B-128E-Instruct", "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "MoE", "context": { "native": 1048576, "url": "https://huggingface.co/unsloth/Llama-4-Maverick-17B-128E-Instruct/resolve/main/config.json", "extended": null }, "license": { "value": "Llama 4 Community License", "url": "https://huggingface.co/meta-llama/Llama-4-Maverick-17B-128E-Instruct/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 49152, "win_per_token": 147456, "win_tokens": 8192, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/unsloth/Llama-4-Maverick-17B-128E-Instruct/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "gemma-3-1b", "name": "Gemma 3 1B IT", "hf_repo": "google/gemma-3-1b-it", "hf_revision": null, "source_url": "https://huggingface.co/google/gemma-3-1b-it", "org": "google", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-03-12", "precision": "day", "url": "https://blog.google/innovation-and-ai/technology/developers-tools/gemma-3/", "status": "verified_primary", "note": null }, "params": { "total_b": 1, "active_b": 1, "moe": false, "url": "https://huggingface.co/api/models/google/gemma-3-1b-it", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 32768, "url": "https://huggingface.co/unsloth/gemma-3-1b-it/resolve/main/config.json", "extended": null }, "license": { "value": "Gemma Terms of Use", "url": "https://ai.google.dev/gemma/terms" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 4096, "win_per_token": 22528, "win_tokens": 512, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/unsloth/gemma-3-1b-it/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "gemma-3-4b", "name": "Gemma 3 4B IT", "hf_repo": "google/gemma-3-4b-it", "hf_revision": null, "source_url": "https://huggingface.co/google/gemma-3-4b-it", "org": "google", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-03-12", "precision": "day", "url": "https://blog.google/innovation-and-ai/technology/developers-tools/gemma-3/", "status": "verified_primary", "note": null }, "params": { "total_b": 4.3, "active_b": 4.3, "moe": false, "url": "https://huggingface.co/api/models/google/gemma-3-4b-it", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/unsloth/gemma-3-4b-it/resolve/main/config.json", "extended": null }, "license": { "value": "Gemma Terms of Use", "url": "https://ai.google.dev/gemma/terms" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 20480, "win_per_token": 118784, "win_tokens": 1024, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/unsloth/gemma-3-4b-it/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "gemma-3-12b", "name": "Gemma 3 12B IT", "hf_repo": "google/gemma-3-12b-it", "hf_revision": null, "source_url": "https://huggingface.co/google/gemma-3-12b-it", "org": "google", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-03-12", "precision": "day", "url": "https://blog.google/innovation-and-ai/technology/developers-tools/gemma-3/", "status": "verified_primary", "note": null }, "params": { "total_b": 12.187, "active_b": 12.187, "moe": false, "url": "https://huggingface.co/api/models/google/gemma-3-12b-it", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/unsloth/gemma-3-12b-it/resolve/main/config.json", "extended": null }, "license": { "value": "Gemma Terms of Use", "url": "https://ai.google.dev/gemma/terms" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 65536, "win_per_token": 327680, "win_tokens": 1024, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/unsloth/gemma-3-12b-it/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "gemma-3-27b", "name": "Gemma 3 27B IT", "hf_repo": "google/gemma-3-27b-it", "hf_revision": null, "source_url": "https://huggingface.co/google/gemma-3-27b-it", "org": "google", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-03-12", "precision": "day", "url": "https://blog.google/innovation-and-ai/technology/developers-tools/gemma-3/", "status": "verified_primary", "note": null }, "params": { "total_b": 27.432, "active_b": 27.432, "moe": false, "url": "https://huggingface.co/api/models/google/gemma-3-27b-it", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/unsloth/gemma-3-27b-it/resolve/main/config.json", "extended": null }, "license": { "value": "Gemma Terms of Use", "url": "https://ai.google.dev/gemma/terms" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 81920, "win_per_token": 425984, "win_tokens": 1024, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/unsloth/gemma-3-27b-it/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "mistral-7b-v0.3", "name": "Mistral 7B Instruct v0.3", "hf_repo": "mistralai/Mistral-7B-Instruct-v0.3", "hf_revision": null, "source_url": "https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.3", "org": "mistralai", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2024-05-22", "precision": "day", "url": "https://docs.mistral.ai/models/mistral-7b-0-3", "status": "verified_primary", "note": null }, "params": { "total_b": 7.248, "active_b": 7.248, "moe": false, "url": "https://huggingface.co/api/models/mistralai/Mistral-7B-Instruct-v0.3", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 32768, "url": "https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.3/resolve/main/config.json", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://www.apache.org/licenses/LICENSE-2.0" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 131072, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.3/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "mistral-nemo-12b", "name": "Mistral NeMo 12B Instruct (2407)", "hf_repo": "mistralai/Mistral-Nemo-Instruct-2407", "hf_revision": null, "source_url": "https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407", "org": "mistralai", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2024-07-18", "precision": "day", "url": "https://mistral.ai/news/mistral-nemo/", "status": "verified_primary", "note": null }, "params": { "total_b": 12.248, "active_b": 12.248, "moe": false, "url": "https://huggingface.co/api/models/mistralai/Mistral-Nemo-Instruct-2407", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407/resolve/main/config.json", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://www.apache.org/licenses/LICENSE-2.0" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 163840, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "mistral-small-3.2-24b", "name": "Mistral Small 3.2 24B Instruct (2506)", "hf_repo": "mistralai/Mistral-Small-3.2-24B-Instruct-2506", "hf_revision": null, "source_url": "https://huggingface.co/mistralai/Mistral-Small-3.2-24B-Instruct-2506", "org": "mistralai", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-06-20", "precision": "day", "url": "https://docs.mistral.ai/models/mistral-small-3-2-25-06", "status": "verified_primary", "note": null }, "params": { "total_b": 24.011, "active_b": 24.011, "moe": false, "url": "https://huggingface.co/api/models/mistralai/Mistral-Small-3.2-24B-Instruct-2506", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/mistralai/Mistral-Small-3.2-24B-Instruct-2506/resolve/main/config.json", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://www.apache.org/licenses/LICENSE-2.0" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 163840, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/mistralai/Mistral-Small-3.2-24B-Instruct-2506/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "mixtral-8x7b", "name": "Mixtral 8x7B Instruct v0.1", "hf_repo": "mistralai/Mixtral-8x7B-Instruct-v0.1", "hf_revision": null, "source_url": "https://huggingface.co/mistralai/Mixtral-8x7B-Instruct-v0.1", "org": "mistralai", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2023-12-11", "precision": "day", "url": "https://mistral.ai/news/mixtral-of-experts/", "status": "verified_primary", "note": "Instruct/public announcement date; do not conflate earlier base-weight torrent with this checkpoint." }, "params": { "total_b": 46.703, "active_b": 12.9, "moe": true, "url": "https://huggingface.co/api/models/mistralai/Mixtral-8x7B-Instruct-v0.1", "active_url": "https://mistral.ai/news/mixtral-of-experts", "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "MoE", "context": { "native": 32768, "url": "https://huggingface.co/mistralai/Mixtral-8x7B-Instruct-v0.1/resolve/main/config.json", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://www.apache.org/licenses/LICENSE-2.0" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 131072, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/mistralai/Mixtral-8x7B-Instruct-v0.1/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "mixtral-8x22b", "name": "Mixtral 8x22B Instruct v0.1", "hf_repo": "mistralai/Mixtral-8x22B-Instruct-v0.1", "hf_revision": null, "source_url": "https://huggingface.co/mistralai/Mixtral-8x22B-Instruct-v0.1", "org": "mistralai", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2024-04-17", "precision": "day", "url": "https://mistral.ai/news/mixtral-8x22b/", "status": "verified_primary", "note": "Instruct/public announcement date." }, "params": { "total_b": 140.63, "active_b": 39, "moe": true, "url": "https://huggingface.co/api/models/mistralai/Mixtral-8x22B-Instruct-v0.1", "active_url": "https://mistral.ai/news/mixtral-8x22b", "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "MoE", "context": { "native": 65536, "url": "https://huggingface.co/mistralai/Mixtral-8x22B-Instruct-v0.1/resolve/main/config.json", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://www.apache.org/licenses/LICENSE-2.0" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 229376, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/mistralai/Mixtral-8x22B-Instruct-v0.1/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "deepseek-r1-distill-qwen-1.5b", "name": "DeepSeek-R1-Distill-Qwen-1.5B", "hf_repo": "deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B", "hf_revision": null, "source_url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B", "org": "deepseek-ai", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-01-20", "precision": "day", "url": "https://api-docs.deepseek.com/news/news250120", "status": "verified_primary", "note": "Original R1 and accompanying distills; not later R1-0528." }, "params": { "total_b": 1.777, "active_b": 1.777, "moe": false, "url": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B/resolve/main/config.json", "extended": null }, "license": { "value": "MIT", "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 28672, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "deepseek-r1-distill-qwen-7b", "name": "DeepSeek-R1-Distill-Qwen-7B", "hf_repo": "deepseek-ai/DeepSeek-R1-Distill-Qwen-7B", "hf_revision": null, "source_url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-7B", "org": "deepseek-ai", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-01-20", "precision": "day", "url": "https://api-docs.deepseek.com/news/news250120", "status": "verified_primary", "note": "Original R1 and accompanying distills; not later R1-0528." }, "params": { "total_b": 7.616, "active_b": 7.616, "moe": false, "url": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-R1-Distill-Qwen-7B", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-7B/resolve/main/config.json", "extended": null }, "license": { "value": "MIT", "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-7B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 57344, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-7B/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "deepseek-r1-distill-llama-8b", "name": "DeepSeek-R1-Distill-Llama-8B", "hf_repo": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B", "hf_revision": null, "source_url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-8B", "org": "deepseek-ai", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-01-20", "precision": "day", "url": "https://api-docs.deepseek.com/news/news250120", "status": "verified_primary", "note": "Original R1 and accompanying distills; not later R1-0528." }, "params": { "total_b": 8.03, "active_b": 8.03, "moe": false, "url": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-R1-Distill-Llama-8B", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-8B/resolve/main/config.json", "extended": null }, "license": { "value": "MIT", "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-8B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 131072, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-8B/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "deepseek-r1-distill-qwen-14b", "name": "DeepSeek-R1-Distill-Qwen-14B", "hf_repo": "deepseek-ai/DeepSeek-R1-Distill-Qwen-14B", "hf_revision": null, "source_url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-14B", "org": "deepseek-ai", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-01-20", "precision": "day", "url": "https://api-docs.deepseek.com/news/news250120", "status": "verified_primary", "note": "Original R1 and accompanying distills; not later R1-0528." }, "params": { "total_b": 14.77, "active_b": 14.77, "moe": false, "url": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-R1-Distill-Qwen-14B", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-14B/resolve/main/config.json", "extended": null }, "license": { "value": "MIT", "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-14B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 196608, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-14B/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "deepseek-r1-distill-qwen-32b", "name": "DeepSeek-R1-Distill-Qwen-32B", "hf_repo": "deepseek-ai/DeepSeek-R1-Distill-Qwen-32B", "hf_revision": null, "source_url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B", "org": "deepseek-ai", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-01-20", "precision": "day", "url": "https://api-docs.deepseek.com/news/news250120", "status": "verified_primary", "note": "Original R1 and accompanying distills; not later R1-0528." }, "params": { "total_b": 32.764, "active_b": 32.764, "moe": false, "url": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B/resolve/main/config.json", "extended": null }, "license": { "value": "MIT", "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 262144, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "deepseek-r1-distill-llama-70b", "name": "DeepSeek-R1-Distill-Llama-70B", "hf_repo": "deepseek-ai/DeepSeek-R1-Distill-Llama-70B", "hf_revision": null, "source_url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-70B", "org": "deepseek-ai", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-01-20", "precision": "day", "url": "https://api-docs.deepseek.com/news/news250120", "status": "verified_primary", "note": "Original R1 and accompanying distills; not later R1-0528." }, "params": { "total_b": 70.554, "active_b": 70.554, "moe": false, "url": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-R1-Distill-Llama-70B", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-70B/resolve/main/config.json", "extended": null }, "license": { "value": "MIT", "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-70B/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 327680, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-70B/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "deepseek-v3", "name": "DeepSeek-V3", "hf_repo": "deepseek-ai/DeepSeek-V3", "hf_revision": null, "source_url": "https://huggingface.co/deepseek-ai/DeepSeek-V3", "org": "deepseek-ai", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2024-12-26", "precision": "day", "url": "https://api-docs.deepseek.com/news/news1226", "status": "verified_primary", "note": "Original V3; not V3-0324 or V3.1." }, "params": { "total_b": 671, "active_b": 37, "moe": true, "url": "https://huggingface.co/deepseek-ai/DeepSeek-V3/raw/main/README.md", "active_url": "https://huggingface.co/deepseek-ai/DeepSeek-V3/raw/main/README.md", "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "MoE", "context": { "native": 163840, "url": "https://huggingface.co/deepseek-ai/DeepSeek-V3/resolve/main/config.json", "extended": null }, "license": { "value": "DeepSeek Model License (code: MIT)", "url": "https://github.com/deepseek-ai/DeepSeek-V3/blob/main/LICENSE-MODEL" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 70272, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/deepseek-ai/DeepSeek-V3/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "deepseek-r1", "name": "DeepSeek-R1", "hf_repo": "deepseek-ai/DeepSeek-R1", "hf_revision": null, "source_url": "https://huggingface.co/deepseek-ai/DeepSeek-R1", "org": "deepseek-ai", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-01-20", "precision": "day", "url": "https://api-docs.deepseek.com/news/news250120", "status": "verified_primary", "note": "Original R1 and accompanying distills; not later R1-0528." }, "params": { "total_b": 671, "active_b": 37, "moe": true, "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1/raw/main/README.md", "active_url": "https://huggingface.co/deepseek-ai/DeepSeek-R1/raw/main/README.md", "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "MoE", "context": { "native": 163840, "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1/resolve/main/config.json", "extended": null }, "license": { "value": "MIT", "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 70272, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/deepseek-ai/DeepSeek-R1/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "gpt-oss-20b", "name": "gpt-oss-20b", "hf_repo": "openai/gpt-oss-20b", "hf_revision": null, "source_url": "https://huggingface.co/openai/gpt-oss-20b", "org": "openai", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-08-05", "precision": "day", "url": "https://openai.com/index/introducing-gpt-oss/", "status": "verified_primary", "note": null }, "params": { "total_b": 20.915, "active_b": 3.6, "moe": true, "url": "https://huggingface.co/api/models/openai/gpt-oss-20b", "active_url": "https://huggingface.co/openai/gpt-oss-20b/raw/main/README.md", "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "MoE", "context": { "native": 131072, "url": "https://huggingface.co/openai/gpt-oss-20b/resolve/main/config.json", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/openai/gpt-oss-20b/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 24576, "win_per_token": 24576, "win_tokens": 128, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/openai/gpt-oss-20b/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "gpt-oss-120b", "name": "gpt-oss-120b", "hf_repo": "openai/gpt-oss-120b", "hf_revision": null, "source_url": "https://huggingface.co/openai/gpt-oss-120b", "org": "openai", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-08-05", "precision": "day", "url": "https://openai.com/index/introducing-gpt-oss/", "status": "verified_primary", "note": null }, "params": { "total_b": 116.829, "active_b": 5.1, "moe": true, "url": "https://huggingface.co/api/models/openai/gpt-oss-120b", "active_url": "https://huggingface.co/openai/gpt-oss-120b/raw/main/README.md", "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "MoE", "context": { "native": 131072, "url": "https://huggingface.co/openai/gpt-oss-120b/resolve/main/config.json", "extended": null }, "license": { "value": "Apache 2.0", "url": "https://huggingface.co/openai/gpt-oss-120b/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 36864, "win_per_token": 36864, "win_tokens": 128, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/openai/gpt-oss-120b/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "phi-4", "name": "Phi-4", "hf_repo": "microsoft/phi-4", "hf_revision": null, "source_url": "https://huggingface.co/microsoft/phi-4", "org": "microsoft", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2024-12-12", "precision": "day", "url": "https://huggingface.co/microsoft/phi-4/raw/main/README.md", "status": "verified_primary", "note": "Publisher-stated initial release; HF availability can be later." }, "params": { "total_b": 14.66, "active_b": 14.66, "moe": false, "url": "https://huggingface.co/api/models/microsoft/phi-4", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 16384, "url": "https://huggingface.co/microsoft/phi-4/resolve/main/config.json", "extended": null }, "license": { "value": "MIT", "url": "https://huggingface.co/microsoft/phi-4/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 204800, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/microsoft/phi-4/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "phi-4-mini", "name": "Phi-4-mini-instruct", "hf_repo": "microsoft/Phi-4-mini-instruct", "hf_revision": null, "source_url": "https://huggingface.co/microsoft/Phi-4-mini-instruct", "org": "microsoft", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-02-26", "precision": "day", "url": "https://aka.ms/phi4-feb2025", "status": "verified_primary", "note": null }, "params": { "total_b": 3.836, "active_b": 3.836, "moe": false, "url": "https://huggingface.co/api/models/microsoft/Phi-4-mini-instruct", "active_url": null, "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "Dense", "context": { "native": 131072, "url": "https://huggingface.co/microsoft/Phi-4-mini-instruct/resolve/main/config.json", "extended": null }, "license": { "value": "MIT", "url": "https://huggingface.co/microsoft/Phi-4-mini-instruct/blob/main/LICENSE" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 131072, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/microsoft/Phi-4-mini-instruct/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] }, { "slug": "glm-4.5-air", "name": "GLM-4.5-Air", "hf_repo": "zai-org/GLM-4.5-Air", "hf_revision": null, "source_url": "https://huggingface.co/zai-org/GLM-4.5-Air", "org": "zai-org", "status": "legacy", "catalog_status": "legacy", "release_2026_verified": false, "default_visible": false, "release": { "label": "2025-07-28", "precision": "day", "url": "https://www.prnewswire.com/news-releases/zai-releases-glm-4-5--setting-new-standards-for-ai-performance-and-accessibility-while-improving-affordability-302514803.html", "status": "verified_primary", "note": "Press release provided by Z.ai and distributed by PRNewswire." }, "params": { "total_b": 110.469, "active_b": 12, "moe": true, "url": "https://huggingface.co/api/models/zai-org/GLM-4.5-Air", "active_url": "https://huggingface.co/zai-org/GLM-4.5-Air/raw/main/README.md", "scope": "v1 database (model card / HF safetensors count)" }, "arch_label": "MoE", "context": { "native": 131072, "url": "https://huggingface.co/zai-org/GLM-4.5-Air/resolve/main/config.json", "extended": null }, "license": { "value": "MIT", "url": "https://opensource.org/license/mit" }, "quality": { "rank": null, "aa": null, "arena": null }, "kv": { "grow": 188416, "win_per_token": 0, "win_tokens": 0, "win_max": null, "fixed": 0, "unknown": [], "status": "estimated", "url": "https://huggingface.co/zai-org/GLM-4.5-Air/blob/main/config.json", "dtype": "FP16", "note": null }, "artifacts": {}, "artifacts_missing": [], "community": [], "notes": [ "Released before 2026: kept for reference, excluded from picks and default rankings. File sizes are estimated from parameter count × bits per weight (no verified artifact bytes in the 2026 research)." ] } ] }