World Models
3 published articles
Computer Vision, World Models & AI Research
PhiZero: teaching video AI to think in physics before it renders
PhiZero, a CASIA world model, learns a compact discrete "physical language" from raw video and uses it to reason about how a scene will evolve before rendering frames. The authors argue this reason-then-render design produces more physically coherent video than direct pixel prediction.
2026-07-31
Embodied AI
Qwen's new robotics models skip the usual AI shortcut
Alibaba's Qwen team released three robotics-specific foundation models and a world model for agent environments, shifting from general-purpose vision-language models to purpose-built architectures for physical action.
2026-07-18
Open Source Robotics
LeRobot v0.6.0 trains robots to see the future, then charges nothing for the vision
LeRobot v0.6.0 introduces world model policies that imagine the future during training and vanish at inference, a new rewards API, six simulation benchmarks, and faster data loading. The framework aims to democratize robot learning with leaner code and cloud training options.
2026-07-11