LocalFit
Submit a benchmark

Falcon-H1-Tiny-90M-Instruct

tiiuae/Falcon-H1-Tiny-90M-Instruct

ReleasedRelease date unknownSize0.1B tensors · denseRoleAssistantLicensefalcon-llm-licenseContext256K config max

Where it runs 8K context · ≥ 5 t/s floor

SetupBest planMemoryGenerate
8 GB GPU + 32 GB RAMFP16llama.cpp CUDAFull GPUKV unknown~892–1,291 t/s Estimated
12 GB GPU + 32 GB RAMFP16llama.cpp CUDAFull GPUKV unknown~1,501–1,601 t/s Estimated
16 GB GPU + 64 GB RAMFP16llama.cpp CUDAFull GPUKV unknown~3,107–3,797 t/s Estimated
24 GB GPU + 64 GB RAMFP16llama.cpp CUDAFull GPUKV unknown~3,138–3,935 t/s Estimated
32 GB GPU + 128 GB RAMFP16llama.cpp CUDAFull GPUKV unknown~4,927–7,868 t/s Estimated
2× 24 GB GPUs + 128 GB RAMFP16llama.cpp CUDAFull GPUKV unknown~2,866–3,700 t/s Estimated
96 GB workstation GPU + 256 GB RAMFP16llama.cpp CUDAFull GPUKV unknown~5,046–7,922 t/s Estimated
Mac, 24 GBFP16llama.cpp MetalUnified memoryKV unknown~1,056–1,260 t/s Estimated
Mac, 48 GBFP16llama.cpp MetalUnified memoryKV unknown~1,056–1,260 t/s Estimated
Mac, 128 GBFP16llama.cpp MetalUnified memoryKV unknown~1,714–2,616 t/s Estimated
Mac, 512 GBFP16llama.cpp MetalUnified memoryKV unknown~1,913–2,942 t/s Estimated
DGX Spark 128 GBFP16llama.cpp CUDAUnified memoryKV unknown~1,132–1,254 t/s Estimated
Ryzen AI Max+ 395, 128 GBFP16llama.cpp VulkanUnified memoryKV unknown~760–1,106 t/s Estimated
CPU only, 64 GB DDR5-5600 dual channelQ4_K_Mllama.cpp CPUCPU onlyKV unknown~347–411 t/s Estimated extrapolated

Same plan engine as the planner, with its default reserves and RAM. How plans are chosen.

Files exact Hugging Face bytes

ArtifactSizeFilesRevisionSource
FP16
mradermacher/Falcon-H1-Tiny-90M-Instruct-GGUF
0.17 GiB1fa186ca32cHF files
Q8_0
mradermacher/Falcon-H1-Tiny-90M-Instruct-GGUF
0.09 GiB1fa186ca32cHF files
Q5_K_M
mradermacher/Falcon-H1-Tiny-90M-Instruct-GGUF
0.06 GiB1fa186ca32cHF files
Q4_K_M
mradermacher/Falcon-H1-Tiny-90M-Instruct-GGUF
0.05 GiB1fa186ca32cHF files
MLX 8-bit 8-bit
mlx-community/Falcon-H1-Tiny-90M-Instruct-8bit
0.09 GiB11c292eaff5HF files
MLX 4-bit 4-bit
mlx-community/Falcon-H1-Tiny-90M-Instruct-4bit
0.05 GiB112db8d70c6HF files
Native safetensors BF16
tiiuae/Falcon-H1-Tiny-90M-Instruct
0.17 GiB1e6389502a0HF files

Not found in the representative quant search (not proven unavailable): EXL3, AWQ, GPTQ. Complete selected file sets; projectors, MTP/draft and lookup files are listed separately and budgeted only by plans that use them. File bytes are not allocated memory.

Quality

No Artificial Analysis evaluation mapped to this checkpoint.

Arena text rating

Unknown

Publisher-reported benchmarks

None extracted from the model card.

AA and Arena describe the hosted model in the stated mode, not each quantized file; scores are never inherited by base, pruned, abliterated or quantized variants and never averaged together.

How to run it

Ollama · Any supportedOllamaCPU/GPU backend; exact memory/speed unknown

publisher-documented, not executed source

ollama run hf.co/tiiuae/Falcon-H1-Tiny-90M-Instruct-GGUF:Q8_0

Prerequisites: See source setup/dependencies; these snippets were inspected, not executed.

  • ExactQ8_0 GGUF for tiny instruction checkpoint.
  • Other card examples contain source typos/wrong model ids (SGLang ttiiuae); those are not reproduced as working recipes.
  • Tiny benchmark or tool competence is not large-model reasoning quality.

Recipes are source examples (settings are the publisher’s or author’s, not measured optima) and were not executed by LocalFit. A downloadable file does not prove your backend build supports the architecture.

Reported speeds 0

No reported measurement yet. Submit one.