Apple M1 Max (32-core GPU) × Meta Llama 3.1 8B Instruct
Meta Llama 3.1 8B Instruct on Apple M1 Max (32-core GPU)
38.4best gen tok/s
392best prompt tok/s
4,352max context tested
10results
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 | 38.4 | 392 | LocalScore ↗ | 512+128 | 3,078 ms | llamafile 0.9.2 |
| Q4_K - Medium | 38.1 | 384 | LocalScore ↗ | 512+128 | 3,116 ms | llamafile 0.9.2 |
| Q4_K - Medium | 37.9 | 385 | LocalScore ↗ | 512+128 | 3,101 ms | llamafile 0.9.2 |
| Q4_K - Medium | 37.7 | 383 | LocalScore ↗ | 512+128 | 3,093 ms | llamafile 0.9.2 |
| Q4_K - Medium | 37.7 | 388 | LocalScore ↗ | 512+128 | 3,113 ms | llamafile 0.9.2 |
| Q4_K - Medium | 37.6 | 389 | LocalScore ↗ | 4,352 | 3,085 ms | llamafile 0.9.2 |
| Q4_K - Medium | 37.5 | 369 | LocalScore ↗ | 512+128 | 3,166 ms | llamafile 0.9.2 |
| Q4_K - Medium | 33.6 | 350 | LocalScore ↗ | 512+128 | 3,540 ms | llamafile 0.9.2 |
| Q4_K - Medium | 32.3 | 338 | LocalScore ↗ | 4,352 | 3,705 ms | llamafile 0.9.2 |
| Q4_K - Medium | 31.6 | 312 | LocalScore ↗ | 512+128 | 4,093 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 M1 Max (32-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.