Hardware /
Quadro RTX 4000
4measured results
67.6best gen tok/s*
8 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 2 7B | Q4_0 | 67.6 | 1,663 | llama.cpp (CUDA) ↗ | 512+128 | llama.cpp CUDA @7d77f07 | Details → |
| Llama 2 7B | Q4_0 | 65.6 | 1,537 | llama.cpp (CUDA) ↗ | 512+128 | llama.cpp CUDA @7d77f07 | Details → |
| Meta Llama 3.1 8B Instruct | Q4_K - Medium | 41.2 | 1,096 | LocalScore ↗ | 512+128 | llamafile 0.9.3 | Details → |
| Meta Llama 3.1 8B Instruct | Q4_K - Medium | 38.8 | 889 | 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.