Alibaba Open-Sources Qwen3.8 Series with 27B Parameter Model Outperforming Predecessor

Alibaba officially open-sourced the Qwen3.8 model series on the evening of August 14, allowing all developers, research institutions, and enterprises to freely download, deploy, and use the models. The newly released Qwen3.8-27B is a native multimodal dense model with 27 billion parameters, delivering overall performance that surpasses its predecessor, Qwen3.7-Plus, particularly in practical scenarios such as coding and office productivity. The model offers fast response times and high output quality, is easy to deploy, and can run smoothly on consumer-grade graphics cards after quantization.

Alibaba has now open-sourced more than 460 models to the global AI community. Total downloads of the Qwen model family have exceeded 3 billion worldwide, with over 300,000 derivative models created, making it the top open-source model family globally. Previously, the flagship Qwen3.8-Max model weights were also made publicly available.