Qwen Releases Qwen3.8-27B-FP8 Vision-Language Weights
Qwen’s Hugging Face repository provides FP8-quantized weights and configuration files for the post-trained Qwen3.8-27B-FP8, a 27B-parameter vision-language model. The artifacts use the Hugging Face Transformers format and are reported as compatible with Transformers, vLLM, SGLang and TokenSpeed. Qwen says fine-grained FP8 quantization with a block size of 128 produces performance nearly identical to the original model.
The release gives developers deployable Apache 2.0-licensed FP8 artifacts for a native vision-language model, potentially simplifying local integration across major inference frameworks. The near-parity performance claim is vendor-reported and lacks independent benchmark evidence in the supplied report.
Qwen/Qwen3.8-27B-FP8 Image-Text-to-Text • 28B • Updated 3 days ago • 5.29M • 980
base_model:
- Qwen/Qwen3.8-27B library_name: transformers license: apache-2.0 pipeline_tag: image-text-to-text
Qwen3.8-27B-FP8
[!Note] This repository contains FP8-quantized model weights and configuration files for the post-trained model in the Hugging Face Transformers format. These artifacts are compatible with Hugging Face Transformers, vLLM, SGLang, TokenSpeed, etc. The quantization method is fine-grained fp8 quantization with block size of 128, and its performance metrics are nearly identical to those of the original model.
#Qwen#Qwen3.8-27B-FP8#Hugging Face Transformers#vLLM#SGLang#TokenSpeed#Qwen Cloud