GPUs & Macs › NVIDIA RTX PRO 6000 Blackwell Server Edition
NVIDIA RTX PRO 6000 Blackwell Server Edition: local LLM speed
2 measured results for the NVIDIA RTX PRO 6000 Blackwell Server Edition.
Measured results
| Model | Generate tok/s | Prompt tok/s | Quant | Backend | Test | Source |
|---|---|---|---|---|---|---|
| Llama 3.2 1B Instruct | 106 | 18,081 | Q4_K_M | llamafile | LocalScore avg | LocalScore details |
| Qwen2.5 14B Instruct | 17.9 | 2,581 | Q4_K_M | llamafile | LocalScore avg | LocalScore details |
What fits on the NVIDIA RTX PRO 6000 Blackwell Server Edition
Estimates for this chip are scaled from a measured run: Qwen2.5 14B Instruct Q4_K_M at 17.9 tok/s (llamafile, source). Fit assumes the whole model in GPU memory with ~8K context.
| Model | Best quant that fits | File | Expected tok/s |
|---|---|---|---|
| Qwen3.8-27B | Q8_0 | 29.0 GB | 5.5 estimate |
| gemma-4-12B-it | Q8_0 | 12.7 GB | 12.7 estimate |
| Qwen3.5-9B | Q8_0 | 9.5 GB | 16.9 estimate |
| Qwen3.5-4B | Q8_0 | 4.5 GB | 36.0 estimate |
| Qwen3.6-35B-A3B | Q8_0 | 36.9 GB | 25.5 estimate |
| Qwen3.8-Flash-Next | UD-IQ3_XXS | 82.0 GB | 2.0 estimate |
| gemma-4-26B-A4B-it | Q8_0 | 26.9 GB | 19.5 estimate |
| MiMo-V2.6-Distill-Qwen-9B | Q8_0 | 9.5 GB | 16.9 estimate |
| gemma-4-E4B-it | Q8_0 | 8.2 GB | 19.7 estimate |
| Muse-Glimmer-30B | Q8_0 | 29.6 GB | 5.4 estimate |
| Qwen3-4B | Q8_0 | 4.3 GB | 37.7 estimate |
| gemma-4-31B-it | Q8_0 | 32.6 GB | 4.9 estimate |
| gemma-4-26B-A4B-it-qat-q4_0-unquantized | UD-Q4_K_XL | 14.2 GB | 36.8 estimate |
| gemma-4-12B-it-qat-q4_0-unquantized | UD-Q4_K_XL | 6.7 GB | 24.0 estimate |
Similar hardware
NVIDIA H100 NVL · NVIDIA H200 · NVIDIA A40-48Q · NVIDIA RTX 5090 · NVIDIA RTX 4090 D · NVIDIA RTX 4090 · NVIDIA RTX 3090 Ti · NVIDIA RTX 3090