Foundational models (FMs) showed comparable performance to ophthalmology experts in textual questions, with Claude 3.5 Sonnet achieving 77.7% accuracy. Multimodal imaging questions revealed ...