On September 9, during the WAVE SUMMIT Deep Learning Developer Conference 2025, Wang Haifeng, Baidu's Chief Technology Officer, formally introduced the Wenxin Large Model X1.1 deep-thinking model. Developed on the foundation of the Wenxin Large Model 4.5, this new model leverages an iterative hybrid reinforcement learning framework, showcasing notable advancements in several key areas, including factual accuracy, adherence to instructions, and agent performance. Specifically, it has realized a 34.8% surge in factual accuracy, a 12.5% uplift in instruction-following capabilities, and a 9.6% enhancement in agent performance. Presently, users can explore this model on the official Wenxin Yiyan website and through the Wenxiaoyan APP. Moreover, the model has been officially rolled out on the Baidu Intelligent Cloud Qianfan Platform, granting comprehensive access to enterprise clients and developers alike.