Our work Towards Practical World Model-based Reinforcement Learning for Vision-Language-Action Models is accepted by ICML 2026!