community pool / model answer
Qwen3.8-Flash-Next-NVFP4
chat · RadixArk/Qwen3.8-Flash-Next-NVFP4
74.5 tok/s · model median · 3 runs in the model record.
| Reference rig | Quantization | Median result | Basis | TTFT (ms) | Peak VRAM | Max context | Engines |
|---|---|---|---|---|---|---|---|
| GB10 Grace Blackwell 128GB | 4-bit | 40.9 tok/s | reported | - | 122.1 GB | 1,048,576 | vllm |
| RTX 5090 32GB | 4-bit | 74.5 tok/s | reported | 77,323 | 31.5 GB | 262,144 | sglang |
| RTX PRO 6000 Blackwell 96GB | 4-bit | 161.2 tok/s | reported | 69 | - GB | 262,144 | 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.