Apple M1 Pro (14-core GPU) × Llama 3.2 1B Instruct
Llama 3.2 1B Instruct on Apple M1 Pro (14-core GPU)
72.9best gen tok/s
1,070best prompt tok/s
4,352max context tested
6results
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 | 72.9 | 1,070 | LocalScore ↗ | 512+128 | 1,118 ms | llamafile 0.9.2 |
| Q4_K - Medium | 72.7 | 1,065 | LocalScore ↗ | 512+128 | 1,125 ms | llamafile 0.9.2 |
| Q4_K - Medium | 72.5 | 1,061 | LocalScore ↗ | 512+128 | 1,125 ms | llamafile 0.9.2 |
| Q4_K - Medium | 72.3 | 1,047 | LocalScore ↗ | 4,352 | 1,142 ms | llamafile 0.9.2 |
| Q4_K - Medium | 71.0 | 1,064 | LocalScore ↗ | 512+128 | 1,134 ms | llamafile 0.9.2 |
| Q4_K - Medium | 69.2 | 964 | LocalScore ↗ | 512+128 | 1,225 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 Pro (14-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.