Apple M3 (10-core GPU) × Llama 3.2 1B Instruct
Llama 3.2 1B Instruct on Apple M3 (10-core GPU)
68.2best gen tok/s
944best prompt tok/s
4,352max context tested
12results
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 | 68.2 | 944 | LocalScore ↗ | 512+128 | 1,404 ms | llamafile 0.9.2 |
| Q4_K - Medium | 67.9 | 939 | LocalScore ↗ | 512+128 | 1,409 ms | llamafile 0.9.2 |
| Q4_K - Medium | 67.8 | 937 | LocalScore ↗ | 512+128 | 1,411 ms | llamafile 0.9.2 |
| Q4_K - Medium | 67.8 | 938 | LocalScore ↗ | 512+128 | 1,405 ms | llamafile 0.9.2 |
| Q4_K - Medium | 67.0 | 927 | LocalScore ↗ | 512+128 | 1,417 ms | llamafile 0.9.2 |
| Q4_K - Medium | 66.8 | 938 | LocalScore ↗ | 4,352 | 1,411 ms | llamafile 0.9.2 |
| Q4_K - Medium | 66.3 | 931 | LocalScore ↗ | 512+128 | 1,390 ms | llamafile 0.9.2 |
| Q4_K - Medium | 65.8 | 906 | LocalScore ↗ | 512+128 | 1,433 ms | llamafile 0.9.2 |
| Q4_K - Medium | 65.6 | 927 | LocalScore ↗ | 512+128 | 1,412 ms | llamafile 0.9.2 |
| Q4_K - Medium | 65.0 | 907 | LocalScore ↗ | 4,352 | 1,464 ms | llamafile 0.9.2 |
| Q4_K - Medium | 63.7 | 917 | LocalScore ↗ | 512+128 | 1,455 ms | llamafile 0.9.2 |
| Q4_K - Medium | 59.5 | 874 | LocalScore ↗ | 512+128 | 1,441 ms | llamafile 0.9.2 |
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 M3 (10-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.