Hardware /
RTX 4070
9measured results
283best gen tok/s*
12 GBreported memory
*Headline excludes tiny/test models (under 0.6B params or TinyStories-style). Open in rig checker →
llama.cpp
| Model | Quant | Gen tok/s | Prompt tok/s | Source | Context tested | Build | |
|---|---|---|---|---|---|---|---|
| Llama 3.2 1B Instruct | Q4_K - Medium | 283 | 12,717 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 2 7B | Q4_0 | 92.3 | 3,179 | llama.cpp (Vulkan) ↗ | 512+128 | llama.cpp Vulkan @9a48399 | Details → |
| Llama 2 7B | Q4_0 | 91.5 | 4,294 | llama.cpp (Vulkan) ↗ | 512+128 | llama.cpp Vulkan @9a48399 — coopmat2 | Details → |
| Meta Llama 3.1 8B Instruct | Q4_K - Medium | 76.3 | 3,192 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 42.6 | 1,752 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Gemma 3 12b It | Q3_K - Large | 36.7 | 1,540 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 36.0 | 1,681 | LocalScore ↗ | 512+128 | llamafile 0.9.3 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 35.0 | 1,642 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Cogito v1 Preview Qwen 14B | Q4_K - Medium | 10.8 | 247 | 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.