On September 9, 2025, at the WAVE SUMMIT Deep Learning Developer Conference, Baidu's CTO, Wang Haifeng, officially introduced the latest iteration of the Wenxin Large Model, X1.1. Developed on the foundation of Wenxin 4.5 training, this model leverages an iterative hybrid reinforcement learning framework, marking several technological advancements. Notably, through the application of reinforcement learning technology incorporating knowledge consistency checks, the model's answer accuracy has surged by 34.8%. Additionally, by utilizing reinforcement learning with instruction validators, its compliance with instructions has seen a 12.5% improvement. Furthermore, employing multi-round reinforcement learning with thought and action chains has enhanced the agent's overall capabilities by 9.6%. Presently, the Wenxin Large Model X1.1 is officially available on Baidu Intelligent Cloud's Qianfan Platform, accessible to enterprise clients and developers alike. Interested users can also explore its features on the official Wenxin Yiyan website and through the Wenxiaoyan APP.