At the media briefing for the unveiling of the Baichuan-M3 model today, Baichuan Intelligence officially launched and open-sourced its cutting-edge medical large language model, Baichuan-M3. This model clinched the top spot globally in the comprehensive evaluation scores of HealthBench—the world's most authoritative medical AI benchmark—and secured first place in the rigorous HealthBench Hard assessment, comprehensively outperforming GPT-5.2. Baichuan-M3 is inherently equipped with "end-to-end" capabilities for handling complex medical consultations, boasting a consultation proficiency that surpasses the average level of human doctors and achieving the lowest hallucination rate in the world. Presently, Baixiaoying, a platform under Baichuan Intelligence, has integrated the M3 model, making its advanced capabilities accessible to both doctors and patients. Wang Xiaochuan, the founder and CEO of Baichuan Intelligence, emphasized that the company's foray into AI-driven medical solutions places a stronger focus on "serious medical" scenarios. He has introduced the "Serious Consultation Paradigm" and the "SCAN Principle" to guide this endeavor. Wang also highlighted the four major challenges currently facing China's medical industry: a scarcity of skilled medical professionals, imbalanced doctor-patient relationships, inadequate primary care services, and underdeveloped medical research. He believes that the most significant growth and service opportunities for AI-powered medical services in the future will primarily lie outside traditional hospital settings, with direct patient engagement being the most effective approach.