Apple M1 (7-core GPU) × Llama 3.2 1B Instruct
Llama 3.2 1B Instruct on Apple M1 (7-core GPU)
38.9best gen tok/s
544best 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 | 38.9 | 537 | LocalScore ↗ | 512+128 | 2,432 ms | llamafile 0.9.2 |
| Q4_K - Medium | 38.6 | 536 | LocalScore ↗ | 512+128 | 2,429 ms | llamafile 0.9.2 |
| Q4_K - Medium | 38.5 | 544 | LocalScore ↗ | 512+128 | 2,390 ms | llamafile 0.9.2 |
| Q4_K - Medium | 38.4 | 540 | LocalScore ↗ | 512+128 | 2,400 ms | llamafile 0.9.2 |
| Q4_K - Medium | 38.4 | 528 | LocalScore ↗ | 512+128 | 2,427 ms | llamafile 0.9.2 |
| Q4_K - Medium | 38.3 | 535 | LocalScore ↗ | 512+128 | 2,430 ms | llamafile 0.9.2 |
| Q4_K - Medium | 37.3 | 509 | LocalScore ↗ | 4,352 | 2,517 ms | llamafile 0.9.2 |
| Q4_K - Medium | 36.0 | 503 | LocalScore ↗ | 512+128 | 2,592 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 (7-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.