On August 11, Baichuan Intelligence proudly introduced Baichuan-M2, an open-source medical-enhanced large model that has achieved a remarkable score of 60.1 on the HealthBench benchmark. Despite its comparatively smaller size of 32 billion parameters, Baichuan-M2 has surpassed OpenAI's latest open-source model, gpt-oss120B, which scored 57.6. This achievement places Baichuan-M2 ahead of all other currently available open-source large models globally, including Qwen3-235B, Deepseek R1, and Kimi K2.