Hardware /
RTX A4500
8measured results
288best gen tok/s*
20 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 | 288 | 10,515 | LocalScore ↗ | 4,352 | llamafile 0.9.2 | Details → |
| Llama 3.2 1B Instruct | Q4_K - Medium | 285 | 10,328 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Llama 2 7B | Q4_0 | 103 | 3,453 | llama.cpp (CUDA) ↗ | 512+128 | llama.cpp CUDA @5cdb27e | Details → |
| Llama 2 7B | Q4_0 | 97.3 | 2,827 | llama.cpp (CUDA) ↗ | 512+128 | llama.cpp CUDA @5cdb27e | Details → |
| Meta Llama 3.1 8B Instruct | Q4_K - Medium | 78.2 | 2,513 | LocalScore ↗ | 4,352 | llamafile 0.9.2 | Details → |
| Meta Llama 3.1 8B Instruct | Q4_K - Medium | 77.6 | 2,484 | LocalScore ↗ | 512+128 | llamafile 0.9.2 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 44.1 | 1,332 | LocalScore ↗ | 4,352 | llamafile 0.9.2 | Details → |
| Qwen2.5 14B Instruct | Q4_K - Medium | 43.8 | 1,330 | 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.