On September 18, 2026, DAMO RADAR, a state-of-the-art general medical imaging AI model, made its debut in the esteemed journal Science. This innovative model was the result of a collaborative effort between Alibaba DAMO Academy and prestigious institutions, including the First Affiliated Hospital of Zhejiang University School of Medicine. Specifically tailored for diagnosing abdominal contrast-enhanced CT scans, DAMO RADAR showcases remarkable capabilities by identifying over 146 types of medical conditions in a single scan, encompassing 18 different organs. Its diagnostic accuracy is on par with that of seasoned expert radiologists. In a significant move towards fostering openness and collaboration in the medical AI community, the model has been made open-source. What sets DAMO RADAR apart is its utilization of a vision-language learning approach, combined with a pioneering 'organ-level fine-grained alignment' strategy. This unique combination allows for scalable, versatile, and interpretable diagnosis without the need for extensive manual annotation. In rigorous testing involving nearly 40,000 real-world examinations, DAMO RADAR achieved an impressive AUC (Area Under the Curve) of 0.913, outperforming an average of 23 doctors in human-machine comparison trials. Moreover, when doctors leveraged the assistance of DAMO RADAR, their diagnostic sensitivity saw a notable 10% improvement, while the time taken for diagnosis was significantly reduced by over 30%. This marks a significant step forward in integrating AI into medical diagnostics, enhancing both efficiency and accuracy.