community pool / model answer
Qwen3.6-27B-NVFP4
chat · nvidia/Qwen3.6-27B-NVFP4
60.1 tok/s · model median · 15 runs in the model record.
| Reference rig | Quantization | Median result | Basis | TTFT (ms) | Peak VRAM | Max context | Engines |
|---|---|---|---|---|---|---|---|
| RTX 5060 Ti 16GB ×2 | 4-bit | 15.9 tok/s | reported | 140 | - GB | 2,048 | vllm |
| RTX 5090 32GB | 4-bit | 55 tok/s | measured | 713 | 27.8 GB | 148,000 | llama.cpp, vllm |
| RTX PRO 6000 Blackwell 96GB | 4-bit | 51.3 tok/s | measured | 65 | - GB | 131,072 | tensorrt-llm, vllm |
| RTX PRO 6000 Blackwell 96GB ×2 | 4-bit | 177.7 tok/s | reported | 46 | - GB | 2,048 | vllm |
| RTX PRO 6000 Blackwell 96GB ×4 | 4-bit | 168.1 tok/s | reported | 52 | - GB | 2,048 | vllm |
Context tested
The table above is the tested context for each exact rig and quantization cell. Empty fields remain empty rather than being filled with an estimate.