Hardware /
Quadro RTX 8000
7measured results
250best gen tok/s*
44 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 | 250 | 9,356 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 3.2 1B Instruct | Q4_K - Medium | 218 | 8,204 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 2 7B | Q4_0 | 104 | 3,052 | llama.cpp (CUDA) ↗ | 512+128 | llama.cpp CUDA @b8e09f0 | Details → |
| Llama 2 7B | Q4_0 | 103 | 2,710 | llama.cpp (CUDA) ↗ | 512+128 | llama.cpp CUDA @b8e09f0 | Details → |
| Meta Llama 3.1 8B Instruct | Q4_K - Medium | 74.0 | 1,677 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 42.9 | 1,178 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 36.7 | 1,087 | 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.