τ0-WM: The Largest Open-Source Embodied World Model Unveiled

A research team from Shanghai Academy of Smart Systems and Agibot has introduced τ0-WM, the world's largest open-source embodied world model with 5 billion parameters and approximately 30,000 hours of pre-training data, including 17,800 hours of real-world remote operation data. With test-time computation, robots can evaluate and refine multiple action candidates before execution.