Our paper “MotionBench: Benchmarking and Improving Fine-grained Video Motion Understanding for Vision Language Models” is accepted by CVPR 2025.