Apple M1 Pro (16-core GPU) × Llama 3.2 1B Instruct
Llama 3.2 1B Instruct on Apple M1 Pro (16-core GPU)
82.3best gen tok/s
1,204best prompt tok/s
4,352max context tested
13results
Measured · all numbers below come from the linked sources
llama.cpp
| Quant | Gen tok/s | Prompt tok/s | Source | Context tested | TTFT | Build |
|---|---|---|---|---|---|---|
| Q4_K - Medium | 82.3 | 1,197 | LocalScore ↗ | 512+128 | 999 ms | llamafile 0.9.2 |
| Q4_K - Medium | 81.1 | 1,198 | LocalScore ↗ | 512+128 | 1,000 ms | llamafile 0.9.2 |
| Q4_K - Medium | 80.5 | 1,204 | LocalScore ↗ | 512+128 | 1,002 ms | llamafile 0.9.0 |
| Q4_K - Medium | 80.0 | 1,192 | LocalScore ↗ | 512+128 | 1,006 ms | llamafile 0.9.2 |
| Q4_K - Medium | 79.1 | 1,177 | LocalScore ↗ | 512+128 | 1,021 ms | llamafile 0.9.2 |
| Q4_K - Medium | 78.8 | 1,204 | LocalScore ↗ | 512+128 | 1,008 ms | llamafile 0.9.2 |
| Q4_K - Medium | 78.7 | 1,150 | LocalScore ↗ | 512+128 | 1,029 ms | llamafile 0.9.2 |
| Q4_K - Medium | 78.5 | 1,193 | LocalScore ↗ | 4,352 | 1,011 ms | llamafile 0.9.2 |
| Q4_K - Medium | 78.1 | 1,190 | LocalScore ↗ | 512+128 | 1,009 ms | llamafile 0.9.2 |
| Q4_K - Medium | 75.6 | 1,122 | LocalScore ↗ | 512+128 | 1,054 ms | llamafile 0.9.2 |
| Q4_K - Medium | 72.5 | 1,078 | LocalScore ↗ | 512+128 | 1,075 ms | llamafile 0.9.2 |
| Q4_K - Medium | 68.6 | 1,068 | LocalScore ↗ | 512+128 | 1,144 ms | llamafile 0.9.2 |
| Q5_K - Medium | 66.8 | 1,079 | LocalScore ↗ | 512+128 | 1,111 ms | llamafile 0.9.3 |
Ollama
No measured Ollama results yet. Submit yours →
MLX
No measured MLX results yet. Submit yours →
"Context tested" is the largest prompt+generation length actually benchmarked (tokens), not a memory-fit estimate.
What else can Apple M1 Pro (16-core GPU) run? →
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.