Apple M5 (10-core GPU) × Llama 3.2 1B Instruct
Llama 3.2 1B Instruct on Apple M5 (10-core GPU)
101best gen tok/s
1,293best prompt tok/s
4,352max context tested
8results
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 | 101 | 1,272 | LocalScore ↗ | 4,352 | 980 ms | llamafile 0.9.2 |
| Q4_K - Medium | 101 | 1,266 | LocalScore ↗ | 4,352 | 981 ms | llamafile 0.9.2 |
| Q4_K - Medium | 101 | 1,293 | LocalScore ↗ | 512+128 | 989 ms | llamafile 0.9.2 |
| Q4_K - Medium | 99.2 | 1,265 | LocalScore ↗ | 4,352 | 984 ms | llamafile 0.9.2 |
| Q4_K - Medium | 99.1 | 1,272 | LocalScore ↗ | 512+128 | 979 ms | llamafile 0.9.2 |
| Q4_K - Medium | 97.8 | 1,283 | LocalScore ↗ | 512+128 | 995 ms | llamafile 0.9.2 |
| Q4_K - Medium | 97.1 | 1,217 | LocalScore ↗ | 512+128 | 1,012 ms | llamafile 0.9.2 |
| Q4_K - Medium | 96.2 | 1,250 | LocalScore ↗ | 512+128 | 990 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 M5 (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.