Models /
DeepSeek R1 Distill Qwen 1.5B
6results
6devices
1.78Bparams
133fastest gen tok/s
llama.cpp
| Hardware | Quant | Gen tok/s | Prompt tok/s | Source | Context tested | Build | |
|---|---|---|---|---|---|---|---|
| Apple M4 Max (32-core GPU) | Q4_K - Medium | 133 | 2,424 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| RTX 4050 Laptop GPU | Q4_K - Medium | 90.3 | 4,795 | LocalScore ↗ | 512+128 | llamafile 0.9.3 | Details → |
| GTX 1060 6GB | Q4_K - Medium | 62.5 | 1,475 | LocalScore ↗ | 512+128 | llamafile 0.9.3 | Details → |
| RTX 3050 | Q4_K - Medium | 60.4 | 2,851 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| GTX 1060 3GB | IQ4_XS - 4.25 bpw | 8.4 | 107 | LocalScore ↗ | 512+128 | llamafile 0.9.3 | Details → |
| GTX 1660 SUPER | IQ4_XS - 4.25 bpw | 7.5 | 164 | LocalScore ↗ | 512+128 | llamafile 0.9.3 | Details → |
Ollama
No measured Ollama results yet. Submit yours →
MLX
No measured MLX results yet. Submit yours →
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.