Full article: Decoupling Visual Parsing and Diagnostic Reasoning for Vision–Language Models (GPT-4o and GPT-5): Analysis Using Thoracic Imaging Quiz Cases

What is the bottleneck in ongoing attempts to use vision-language models to interpret radiologic imaging? Pranjal Rai, MD, discusses this recent AJR model by Han et al. that seeks to differentiate the roles of visual parsing and diagnostic reasoning toward impacting VLM performance.