Qwen3.8-27B open weights released, tops Hugging Face

Qwen3.8-27B, a native multimodal dense model with 27B parameters, outperforms Qwen3.7-Plus overall and shines in coding & office workflows. It has 262K native context, extendable to 1M. Community releases include uncensored and GGUF versions.