Safetensors NVFP4
Community buildA quantized build of the model Qwen3.8-27B.
- Publisher
- RadixArk
- Source platform
- HUGGINGFACE
- Quantization
- NVFP4
- Revision
- 319f741cce68d7914884900c138a1fbb70a42f30
- Updated
- 2026-09-26
- Total size
- 20.42 GB
File list
Paths and sizes come from the release page; a missing size shows "not provided".
| File path | Size |
|---|---|
| model-00001-of-00003.safetensors | 9.28 GB |
| model-00002-of-00003.safetensors | 9.3 GB |
| model-00003-of-00003.safetensors | 1.83 GB |
Linked benchmarks
Only benchmarks tied to this exact release are listed.
| Hardware | Framework | Quant | Decode | VRAM used | Samples | Evidence |
|---|---|---|---|---|---|---|
| NVIDIA RTX PRO 6000 Blackwell | SGLang | NVFP4 | 171.1 tok/s | 88.32 GB | 1 | L0 Self-reported |
Deployment guidance
This site doesn't guess whether or how this build runs on your machine. General local-deployment tutorials are linked below; if a link or data on this page is wrong, please tell us.