Hardware /
RTX 4070 SUPER
12measured results
238best 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 | 238 | 13,799 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 3.2 1B Instruct | Q4_K - Medium | 218 | 13,459 | LocalScore ↗ | 512+128 | llamafile 0.9.3 | Details → |
| Llama 3.2 1B Instruct | Q4_K - Medium | 138 | 12,894 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 3.2 1B Instruct | Q4_K - Medium | 137 | 12,969 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 2 7B | Q4_0 | 112 | 5,199 | llama.cpp (Vulkan) ↗ | 512+128 | llama.cpp Vulkan @c945aaa — coopmat2 | Details → |
| Llama 2 7B | Q4_0 | 109 | 4,608 | llama.cpp (Vulkan) ↗ | 512+128 | llama.cpp Vulkan @c945aaa — coopmat2 | Details → |
| Meta Llama 3.1 8B Instruct | Q4_K - Medium | 68.0 | 3,682 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Meta Llama 3.1 8B Instruct | Q4_K - Medium | 45.5 | 3,371 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 38.8 | 2,012 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 24.9 | 1,793 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Meta Llama 3.1 8B Instruct | Q4_K - Medium | 22.6 | 2,594 | LocalScore ↗ | 4,352 | llamafile 0.9.3 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 10.9 | 1,299 | 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.