Models /
LLaMA v2
3results
3devices
6.74Bparams
62.5fastest gen tok/s
llama.cpp
| Hardware | Quant | Gen tok/s | Prompt tok/s | Source | Context tested | Build | |
|---|---|---|---|---|---|---|---|
| Apple M4 Max (40-core GPU) | Q4_0 | 62.5 | 692 | LocalScore ↗ | 512+128 | llamafile 0.9.3 | Details → |
| Tesla P40 | Q4_0 | 40.9 | 833 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Apple M4 Pro (16-core GPU) | Q8_0 | 25.5 | 335 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
Ollama
No measured Ollama results yet. Submit yours →
MLX
No measured MLX results yet. Submit yours →
Includes results from LocalScore, a Mozilla Builders project by cjpais, mirrored in the cn0303/local-llm-speed-benchmarks dataset. LocalScore runs use llamafile, which is built on llama.cpp.