Study shows two ChatGPT large language models (LLMs) outperformed human neurology students in an entrance exam, with one model achieving 85% accuracy, demonstrating their potential in clinical neurology and medical education.