On the evening of August 14, Ali Qianwen officially open-sourced the Qwen3.8 series model. All developers, scientific research institutions and enterprises can freely download, deploy and use it. The new open source Qwen 3.8-27B is a native multi-modal dense model. With only 27 billion parameters, the overall level surpasses Qwen 3.7-Plus, and performs well in real programming and office scenarios. The new model responds quickly, has high completion quality, and is easy to deploy. After quantification, it can run smoothly on “consumer-grade” video cards. It is the open source model most anticipated by the global AI community. Qwen3.8-27B is open source using the loose Apache 2.0 protocol and is expected to become the main model for local deployment.

On the evening of August 14, Ali Qianwen officially open-sourced the Qwen3.8 series model. All developers, scientific research institutions and enterprises can freely download, deploy and use it. The new open source Qwen 3.8-27B is a native multi-modal dense model. With only 27 billion parameters, the overall level surpasses Qwen 3.7-Plus, and performs well in real programming and office scenarios. The new model responds quickly, has high completion quality, and is easy to deploy. After quantification, it can run smoothly on “consumer-grade” video cards. It is the open source model most anticipated by the global AI community. Qwen3.8-27B is open source using the loose Apache 2.0 protocol and is expected to become the main model for local deployment.