The explosive progress of large language models (LLMs) over the last three years has reshaped knowledge-intensive disciplines, including medicine and, increasingly, radiology [1,2]. State-of-the-art proprietary systems such as OpenAI GPT-4 have matched—or even exceeded—human pass rates on medical licensing and specialty examinations, demonstrating that transformer-based models can internalize vast biomedical corpora and recall them with examination-level precision.