Alibaba Unveils Massive Qwen3.8-Max AI Model with 2.4T Parameters
Alibaba introduced the Qwen3.8-Max, a 2.4-trillion-parameter AI model that uses only 95 billion parameters per query via a mixture-of-experts architecture, cutting compute costs. The model handles text, images, and video with a 1M-token context window and will have its weights open-sourced next week.
Alibaba has launched Qwen3.8-Max, the largest model in its Qwen family. Despite its total size of 2.4 trillion parameters, the model activates just 95 billion parameters for each inference by employing a mixture-of-experts design. This approach reduces computational overhead by only engaging the components necessary for a given task.
The model is multimodal, processing text, images, and video, and supports a context window of up to 1 million tokens. That capacity enables analysis of lengthy documents, codebases, and extended recordings.
In external benchmarks, Qwen3.8-Max became the top Chinese model for text on Arena.AI and placed second globally for image analysis. Alibaba plans to release its weights as open source within the next week.
Source: ForkLog