Performance of generative large language models in answering questions from the Brazilian Retina and Vitreous Society certification exam.

Key Points

• To evaluate the performance of various generative language models in answering certification exam questions for retina specialists.
• Assessed Claude 3.5 Sonnet, ChatGPT-4, and ChatGPT-4o on certification exam questions from the Brazilian Retina and Vitreous Society.
• Compared performance metrics of the models in the context of medical education.
• Claude 3.5 Sonnet and ChatGPT-4o significantly outperformed ChatGPT-4.
• Enhanced functionalities of the top models indicate their effectiveness for studying board exams.