Models /
Gemma 3 4b It
8results
6devices
4.55Bparams
74.4fastest gen tok/s
llama.cpp
| Hardware | Quant | Gen tok/s | Prompt tok/s | Source | Context tested | Build | |
|---|---|---|---|---|---|---|---|
| RTX 4060 Ti | Q4_K - Medium | 74.4 | 4,885 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| GTX 1080 Ti | Q4_K - Medium | 57.4 | 1,689 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| GTX 1080 Ti | Q4_K - Medium | 57.3 | 1,628 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| RTX 4050 Laptop GPU | Q4_K - Medium | 46.7 | 2,551 | LocalScore ↗ | 512+128 | llamafile 0.9.3 | Details → |
| RTX 4050 Laptop GPU | Q4_K - Medium | 45.5 | 2,455 | LocalScore ↗ | 512+128 | llamafile 0.9.3 | Details → |
| RTX 3050 | Q4_K - Medium | 37.0 | 1,421 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| RTX 2060 | Q6_K | 24.9 | 716 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| RTX 3060 | Q6_K | 23.2 | 1,063 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | 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.