Purpose/Significance Starting from DeepSeek, the paper explores the technological innovation and challenges of medical large language models, in order to address the unique technical difficulties and challenges in the medical field. Method/Process It focuses on analyzing the breakthrough path of medical artificial intelligence(AI) in multimodal data fusion, adaptation of doctors’ thinking patterns, and technical requirements for high-risk medical decision-making, and looks forward the construction of the smart hospitals and the reconstruction of medical service modes using medical large language models in the future.Result/Conclusion With the continuous development of AI technology, medical large language models will play an important role in future medical practice. By integrating multimodal data, introducing causal reasoning, and improving model interpretability, large language models can not only enhance diagnostic efficiency and accuracy, but also provide strong technical support for the construction of digital and smart hospitals and the reconstruction of medical service modes.