RTX PRO 6000 Blackwell × Qwen2.5 14B Instruct
Qwen2.5 14B Instruct on RTX PRO 6000 Blackwell
84.8best gen tok/s
5,568best prompt tok/s
640max context tested
5results
Measured · all numbers below come from the linked sources
llama.cpp
| Quant | Gen tok/s | Prompt tok/s | Source | Context tested | TTFT | Build |
|---|---|---|---|---|---|---|
| Q4_K - Medium | 84.8 | 5,568 | LocalScore ↗ | 512+128 | 229 ms | llamafile 0.9.2 |
| Q4_K - Medium | 81.1 | 5,126 | LocalScore ↗ | 512+128 | 245 ms | llamafile 0.9.2 |
| Q4_K - Medium | 79.7 | 5,072 | LocalScore ↗ | 512+128 | 246 ms | llamafile 0.9.2 |
| Q4_K - Medium | 21.5 | 3,215 | LocalScore ↗ | 512+128 | 396 ms | llamafile 0.9.2 |
| Q4_K - Medium | 14.3 | 1,947 | LocalScore ↗ | 512+128 | 878 ms | llamafile 0.9.2 |
Ollama
No measured Ollama results yet. Submit yours →
MLX
No measured MLX results yet. Submit yours →
"Context tested" is the largest prompt+generation length actually benchmarked (tokens), not a memory-fit estimate.
What else can RTX PRO 6000 Blackwell run? →
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.