Alibaba has launched Qwen3.8-Max, its most powerful AI model yet, featuring 2.4 trillion parameters and a million-token context window.
Alibaba has introduced its newest AI model, Qwen3.8-Max, claiming it to be the most powerful in the Qwen series. The model features an impressive 2.4 trillion parameters and supports a context window of 1 million tokens, making it suitable for advanced programming, practical applications, in-depth research, and long-term challenges.
Developers have implemented a sparse architecture called Mixture-of-Experts, allowing the model to efficiently manage tasks by activating only specialized components as needed, rather than the entire system for each request.
Unlike previous flagship models, the weights for Qwen3.8-Max will be publicly accessible, marking a shift from the proprietary approach of earlier models that were limited to Alibaba's cloud services.
On Arena.AI, Qwen3.8-Max has achieved recognition as the top Chinese text model, ranking fourth overall, just below Claude Fable 5 and the Claude Opus variations by Anthropic. It scored 1,668 points, closely trailing Claude Opus 5 (Max) at 1,705 points and Kimi K3 (Max) at 1,676 points. In the visual ranking, it holds the second position globally.
This launch comes amidst fierce competition among Chinese AI developers, with Moonshot AI recently unveiling the Kimi K3 model, which features 2.8 trillion parameters and native image processing capabilities.
Prior to this, Alibaba had released Qwen3.7-Max, targeting AI agents, and a suite of models for robotics, known as the Qwen-Robot Suite, which includes various specialized functions for navigation and object manipulation. Additionally, in January, Alibaba Cloud's Qwen-3 became the first AI model to be uploaded and operated in orbit.
