PhysBench: Benchmarking and Enhancing Vision-Language Models for Physical World UnderstandingShare on Twitter Facebook LinkedIn Previous Next