【#Tech24H】On January 13, Baichuan Intelligence officially open-sourced its new-generation medical large language model, Baichuan-M3. The model achieved a top global ranking with a comprehensive score of 65.1 on HealthBench, the world’s most authoritative medical AI evaluation platform. It also claimed first place on HealthBench Hard, which specifically tests complex decision-making abilities, with a score of 44.4. Meanwhile, Baichuan-M3 possesses native “end-to-end” capabilities for serious medical consultations, enabling it to proactively ask follow-up questions, gradually narrow down issues, extract key medical history and risk signals, and conduct in-depth medical reasoning based on complete information. Currently, the medical application “Baixiaoying” has integrated M3, making its capabilities available to both doctors and patients.