Apple M4 (10-core GPU) × Qwen2.5 14B Instruct
Qwen2.5 14B Instruct on Apple M4 (10-core GPU)
9.8best gen tok/s
96.3best prompt tok/s
4,352max context tested
12results
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 | 9.8 | 94.2 | LocalScore ↗ | 512+128 | 13,170 ms | llamafile 0.9.2 |
| Q4_K - Medium | 9.6 | 96.3 | LocalScore ↗ | 512+128 | 13,304 ms | llamafile 0.9.2 |
| Q4_K - Medium | 9.6 | 95.8 | LocalScore ↗ | 512+128 | 13,391 ms | llamafile 0.9.2 |
| Q4_K - Medium | 9.6 | 94.3 | LocalScore ↗ | 4,352 | 13,622 ms | llamafile 0.9.2 |
| Q4_K - Medium | 9.5 | 91.6 | LocalScore ↗ | 4,352 | 13,889 ms | llamafile 0.9.2 |
| Q4_K - Medium | 9.4 | 93.7 | LocalScore ↗ | 4,352 | 13,734 ms | llamafile 0.9.2 |
| Q4_K - Medium | 9.2 | 93.3 | LocalScore ↗ | 512+128 | 13,715 ms | llamafile 0.9.2 |
| Q4_K - Medium | 9.2 | 90.4 | LocalScore ↗ | 512+128 | 14,524 ms | llamafile 0.9.2 |
| Q4_K - Medium | 8.5 | 73.6 | LocalScore ↗ | 512+128 | 16,808 ms | llamafile 0.9.2 |
| Q4_K - Medium | 8.1 | 71.2 | LocalScore ↗ | 512+128 | 16,850 ms | llamafile 0.9.2 |
| Q4_K - Medium | 8.0 | 70.5 | LocalScore ↗ | 512+128 | 17,301 ms | llamafile 0.9.2 |
| Q4_K - Medium | 6.5 | 61.3 | LocalScore ↗ | 512+128 | 20,420 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.