LMOD introduces the first comprehensive multimodal ophthalmology dataset designed to benchmark and improve large vision-language models for clinical eye care applications.

• Creates a specialized benchmark spanning multiple eye imaging modalities
• Evaluates current LVLMs' capability to interpret complex ophthalmic imagery
• Establishes foundation for AI systems that can assist in diagnosis and treatment planning
• Addresses critical need for specialized medical AI training resources

This research matters because it could significantly expand access to quality eye care globally, reduce diagnostic delays, and support clinicians in managing increasing patient loads — ultimately helping prevent avoidable vision loss through earlier intervention.

LMOD: A Large Multimodal Ophthalmology Dataset and Benchmark for Large Vision-Language Models