Mellum2-12B-A2.5B-Thinking Coding
Where it runs 8K context · ≥ 5 t/s floor
| Setup | Best plan | Memory | Generate |
|---|---|---|---|
| 8 GB GPU + 32 GB RAM | Q8_0llama.cpp CUDAGPU + CPU experts · 15 expert layers on CPU | Fits | ~7.6–12 t/s Estimated extrapolated |
| 12 GB GPU + 32 GB RAM | Q8_0llama.cpp CUDAGPU + CPU experts · 5 expert layers on CPU | Fits | ~11–19 t/s Estimated extrapolated |
| 16 GB GPU + 64 GB RAM | Q8_0llama.cpp CUDAFull GPU | Fits | ~63–181 t/s Estimated extrapolated |
| 24 GB GPU + 64 GB RAM | Q8_0llama.cpp CUDAFull GPU | Fits | ~71–204 t/s Estimated extrapolated |
| 32 GB GPU + 128 GB RAM | Q8_0llama.cpp CUDAFull GPU | Fits | ~125–363 t/s Estimated extrapolated |
| 2× 24 GB GPUs + 128 GB RAM | Q8_0llama.cpp CUDAFull GPU | Fits | ~66–189 t/s Estimated extrapolated |
| 96 GB workstation GPU + 256 GB RAM | Q8_0llama.cpp CUDAFull GPU | Fits | ~125–363 t/s Estimated extrapolated |
| Mac, 24 GB | MLX 8-bitMLX-LMUnified memory | Fits | ~29–58 t/s Estimated |
| Mac, 48 GB | MLX 8-bitMLX-LMUnified memory | Fits | ~29–58 t/s Estimated |
| Mac, 128 GB | MLX 8-bitMLX-LMUnified memory | Fits | ~63–83 t/s Estimated |
| Mac, 512 GB | MLX 8-bitMLX-LMUnified memory | Fits | ~58–163 t/s Estimated |
| DGX Spark 128 GB | Q8_0llama.cpp CUDAUnified memory | Fits | ~30–82 t/s Estimated |
| Ryzen AI Max+ 395, 128 GB | Q8_0llama.cpp VulkanUnified memory | Fits | ~18–52 t/s Estimated extrapolated |
| CPU only, 64 GB DDR5-5600 dual channel | Q8_0llama.cpp CPUCPU only | Fits | ~7.5–8.9 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
| Artifact | Size | Files | Revision | Source |
|---|---|---|---|---|
| Q8_0 JetBrains/Mellum2-12B-A2.5B-Thinking-GGUF-Q8_0 | 12.04 GiB | 1 | 6f5b0031c9 | HF files |
| MLX 8-bit 8-bit mlx-community/Mellum2-12B-A2.5B-Thinking-8bit | 12.02 GiB | 3 | 9f7ad1338d | HF files |
| MLX 4-bit 4-bit mlx-community/Mellum2-12B-A2.5B-Thinking-4bit | 6.37 GiB | 2 | 0821bd05a5 | HF files |
| EXL3 4-bit blockblockblock/Mellum2-12B-A2.5B-Thinking-exl3-4.0bpw | 6.12 GiB | 1 | b6a84ebe47 | HF files |
| Native safetensors BF16 JetBrains/Mellum2-12B-A2.5B-Thinking | 22.63 GiB | 5 | a731155055 | HF files |
Not found in the representative quant search (not proven unavailable): GGUF FP16, GGUF Q5_K_M, GGUF Q4_K_M, 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
No model-specific recipe in the research. The planner binds the backend’s documented command template (llama.cpp, Ollama, LM Studio, MLX, vLLM) to your plan: open a plan’s details.
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.