Apple M4 (10-core GPU) × Llama 3.2 1B Instruct
Llama 3.2 1B Instruct on Apple M4 (10-core GPU)
78.6best gen tok/s
1,141best prompt tok/s
4,352max context tested
24results
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 | 78.6 | 1,096 | LocalScore ↗ | 512+128 | 1,165 ms | llamafile 0.9.2 |
| Q4_K - Medium | 78.4 | 1,141 | LocalScore ↗ | 512+128 | 1,158 ms | llamafile 0.9.2 |
| Q4_K - Medium | 78.1 | 1,126 | LocalScore ↗ | 512+128 | 1,160 ms | llamafile 0.9.2 |
| Q4_K - Medium | 78.0 | 1,134 | LocalScore ↗ | 512+128 | 1,156 ms | llamafile 0.9.2 |
| Q4_K - Medium | 77.7 | 1,104 | LocalScore ↗ | 512+128 | 1,176 ms | llamafile 0.9.2 |
| Q4_K - Medium | 77.2 | 1,086 | LocalScore ↗ | 4,352 | 1,184 ms | llamafile 0.9.2 |
| Q4_K - Medium | 77.1 | 1,111 | LocalScore ↗ | 512+128 | 1,179 ms | llamafile 0.9.2 |
| Q4_K - Medium | 76.9 | 1,097 | LocalScore ↗ | 512+128 | 1,184 ms | llamafile 0.9.2 |
| Q4_K - Medium | 76.6 | 1,110 | LocalScore ↗ | 512+128 | 1,184 ms | llamafile 0.9.2 |
| Q4_K - Medium | 76.6 | 1,105 | LocalScore ↗ | 512+128 | 1,177 ms | llamafile 0.9.2 |
| Q4_K - Medium | 76.6 | 1,101 | LocalScore ↗ | 4,352 | 1,193 ms | llamafile 0.9.2 |
| Q4_K - Medium | 76.4 | 1,103 | LocalScore ↗ | 512+128 | 1,185 ms | llamafile 0.9.2 |
| Q4_K - Medium | 76.0 | 1,114 | LocalScore ↗ | 512+128 | 1,179 ms | llamafile 0.9.2 |
| Q4_K - Medium | 75.9 | 1,116 | LocalScore ↗ | 512+128 | 1,187 ms | llamafile 0.9.2 |
| Q4_K - Medium | 75.2 | 1,100 | LocalScore ↗ | 512+128 | 1,204 ms | llamafile 0.9.2 |
| Q4_K - Medium | 75.2 | 1,113 | LocalScore ↗ | 512+128 | 1,191 ms | llamafile 0.9.2 |
| Q4_K - Medium | 74.6 | 1,009 | LocalScore ↗ | 512+128 | 1,293 ms | llamafile 0.9.2 |
| Q4_K - Medium | 74.5 | 1,064 | LocalScore ↗ | 512+128 | 1,214 ms | llamafile 0.9.2 |
| Q4_K - Medium | 74.4 | 982 | LocalScore ↗ | 512+128 | 1,303 ms | llamafile 0.9.2 |
| Q4_K - Medium | 74.0 | 1,020 | LocalScore ↗ | 4,352 | 1,292 ms | llamafile 0.9.2 |
| Q4_K - Medium | 74.0 | 992 | LocalScore ↗ | 512+128 | 1,299 ms | llamafile 0.9.2 |
| Q4_K - Medium | 73.4 | 1,008 | LocalScore ↗ | 4,352 | 1,298 ms | llamafile 0.9.2 |
| Q4_K - Medium | 72.5 | 950 | LocalScore ↗ | 512+128 | 1,318 ms | llamafile 0.9.2 |
| Q4_K - Medium | 66.3 | 855 | LocalScore ↗ | 4,352 | 1,540 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 M4 (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.