Hardware /
Apple M2 Max (38-core GPU)
12measured results
153best gen tok/s*
32 GBreported memory
*Headline excludes tiny/test models (under 0.6B params or TinyStories-style). Open in rig checker →
llama.cpp
| Model | Quant | Gen tok/s | Prompt tok/s | Source | Context tested | Build | |
|---|---|---|---|---|---|---|---|
| Llama 3.2 1B Instruct | Q4_K - Medium | 153 | 2,551 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 3.2 1B Instruct | Q4_K - Medium | 148 | 2,494 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 2 7B | Q4_0 | 66.0 | 671 | llama.cpp (Metal) ↗ | 512+128 | llama.cpp Metal | Details → |
| Meta Llama 3.1 8B Instruct | Q4_K - Medium | 46.4 | 484 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Meta Llama 3.1 8B Instruct | Q4_K - Medium | 44.7 | 474 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 2 7B | Q8_0 | 41.8 | 678 | llama.cpp (Metal) ↗ | 512+128 | llama.cpp Metal | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 25.2 | 252 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 2 7B | F16 | 24.6 | 756 | llama.cpp (Metal) ↗ | 512+128 | llama.cpp Metal | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 23.8 | 246 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 23.4 | 232 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 20.6 | 226 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 18.8 | 191 | 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.