On November 16, Alibaba unveiled Qwen4-Max, its new-generation foundation model. The model features 2.4 trillion total parameters, native visual capabilities, and support for million-level context. Powered by an innovative architecture and training optimizations, it delivers stronger performance with lower token consumption and continues to advance in key areas such as agent capabilities and coding. It ranks sixth on CodeArena, the global front-end coding blind test, and sits in the leading tier in third-party evaluations of agent capabilities and professional tasks, including Agents' Last Exam (ALE) and the Vals Index. Qwen4 is deeply optimized for real-world tasks across domains, enabling more efficient invocation of professional tools to handle long-cycle tasks, and it is accelerating AI deployment in professional fields such as automobile manufacturing, game development, financial investment research, and scientific research.
