The Institute of Automation at the Chinese Academy of Sciences has recently unveiled the MV-MATH dataset, a groundbreaking resource for evaluating the mathematical reasoning capabilities of Multi-modal Large Language Models (MLLMs) in complex visual scenarios. This innovative dataset has been accepted by CVPR 2025, solidifying its significance in the field. MV-MATH presents a formidable challenge to MLLMs, as experimental results reveal that even advanced models like GPT-4o struggle to perform well on this benchmark. The dataset comprises 2009 mathematical problems, each integrating multiple images and texts, spanning across 11 distinct mathematical domains. This comprehensive approach ensures a rigorous assessment of the reasoning abilities of MLLMs.
