community pool / model answer

Qwen3.6-27B-Text-NVFP4-MTP

chat · sakamakismile/Qwen3.6-27B-Text-NVFP4-MTP

95.7 tok/s · model median · 6 runs in the model record.

Reference rigQuantizationMedian resultBasisTTFT (ms)Peak VRAMMax contextEngines
RTX 5090 32GB4-bit121.9 tok/sreported13130 GB131,072vllm
RTX 5090 32GB ×24-bit85.3 tok/sreported2,75259.3 GB262,144vllm
RTX PRO 6000 Blackwell 96GB4-bit117.4 tok/sreported75- GB262,144vllm

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.