Research is being conducted on the performance of large language models, including various versions of GPT, on medical board exams. The studies appear to be evaluating the accuracy and clinical reasoning of these models in different medical specialties and exams, including rheumatology and internal medicine. The specific models and exams being compared vary across the studies, with some focusing on GPT-4 and GPT-5, while others include additional models like Gemini and Google Bard.