Can AI speak endo? A multi-platform evaluation of large language models against ESHRE endometriosis guidelines

The authors aimed to evaluate the performance of three widely accessible large language models in accurately, consistently, and reliably answering clinically relevant questions based on the 2022 ESHRE guidelines for endometriosis. They found that while one model (Model A) achieved the highest accuracy and another (Model C) demonstrated superior consistency, all models exhibited suboptimal reliability under free-tier access, highlighting the need for expert oversight in clinical applications. The study provides a benchmark for assessing AI tools in reproductive medicine and suggests that model performance may vary significantly based on access tier.