ReXInTheWild: A Unified Benchmark for Medical Photograph Understanding
It addresses the problem of assessing AI models for medical image interpretation in telemedicine, providing a clinically grounded benchmark, though it is incremental as it focuses on evaluation rather than new methods.
The paper tackles the lack of a comprehensive benchmark for evaluating vision-language models on medical photographs, introducing ReXInTheWild with 955 clinician-verified questions across 484 images, where leading models like Gemini-3 achieve 78% accuracy while MedGemma scores only 37%.
Everyday photographs taken with ordinary cameras are already widely used in telemedicine and other online health conversations, yet no comprehensive benchmark evaluates whether vision-language models can interpret their medical content. Analyzing these images requires both fine-grained natural image understanding and domain-specific medical reasoning, a combination that challenges both general-purpose and specialized models. We introduce ReXInTheWild, a benchmark of 955 clinician-verified multiple-choice questions spanning seven clinical topics across 484 photographs sourced from the biomedical literature. When evaluated on ReXInTheWild, leading multimodal large language models show substantial performance variation: Gemini-3 achieves 78% accuracy, followed by Claude Opus 4.5 (72%) and GPT-5 (68%), while the medical specialist model MedGemma reaches only 37%. A systematic error analysis also reveals four categories of common errors, ranging from low-level geometric errors to high-level reasoning failures and requiring different mitigation strategies. ReXInTheWild provides a challenging, clinically grounded benchmark at the intersection of natural image understanding and medical reasoning. The dataset is available on HuggingFace.