The Tongyi Qianwen team under Alibaba Cloud released an early preview of the Qwen3-Max-Thinking inference model. This version of the model achieved 100% accuracy in the AIME25 and HMMT25 mathematical reasoning benchmark tests and is undergoing sequential training and optimization.Related NewsCICC Lists Predicted Capital Flows for Stocks w/ Big Weighting Changes in MSCI China Index Quarterly Review (Table)Tongyi Qianwen revealed that the Qwen3-Max series will focus on complex logic and mathematical tasks, with subsequent iterative updates aimed at further enhancing generalization capabilities.