CoWorld Reasoner Uses Video World Models for Spatiotemporal Reasoning

CoWorld Reasoner introduces an agentic framework that employs Video World Models (VWMs) as a visual reasoning engine. The system uses a VWM-driven harness to allow Vision-Language Models to simulate visual evidence and perform flexible spatiotemporal reasoning.