Visual Generation Unlocks Human-Like Reasoning through Multimodal World Models Paper • 2601.19834 • Published 1 day ago • 22