Hardware /
Apple M3 Max (40-core GPU)
20measured results
153best gen tok/s*
48 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 | 3,348 | LocalScore ↗ | 4,352 | llamafile 0.9.2 | Details → |
| Llama 3.2 1B Instruct | Q4_K - Medium | 150 | 3,395 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 3.2 1B Instruct | Q4_K - Medium | 148 | 3,403 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 3.2 1B Instruct | Q4_K - Medium | 148 | 3,209 | LocalScore ↗ | 4,352 | llamafile 0.9.2 | Details → |
| Llama 3.2 1B Instruct | Q4_K - Medium | 146 | 3,292 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 3.2 1B Instruct | Q4_K - Medium | 109 | 2,585 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 3.2 1B Instruct | Q4_K - Medium | 105 | 2,458 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 2 7B | Q4_0 | 66.3 | 760 | llama.cpp (Metal) ↗ | 512+128 | llama.cpp Metal | Details → |
| Meta Llama 3.1 8B Instruct | Q4_K - Medium | 46.0 | 583 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Meta Llama 3.1 8B Instruct | Q4_K - Medium | 45.6 | 591 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 2 7B | Q8_0 | 42.8 | 758 | llama.cpp (Metal) ↗ | 512+128 | llama.cpp Metal | Details → |
| Meta Llama 3.1 8B Instruct | Q4_K - Medium | 25.8 | 383 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 25.5 | 303 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 25.5 | 302 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Meta Llama 3.1 8B Instruct | Q4_K - Medium | 25.1 | 371 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 2 7B | F16 | 25.1 | 779 | llama.cpp (Metal) ↗ | 512+128 | llama.cpp Metal | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 24.0 | 296 | LocalScore ↗ | 4,352 | llamafile 0.9.2 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 23.6 | 287 | LocalScore ↗ | 4,352 | llamafile 0.9.2 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 13.9 | 202 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 13.7 | 200 | 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.