A large scale language model based on discrete-state diffusion, specialized for code generation, achieves an inference speed of 2,146 token/s, a 5.4x improvement over autoregressive models of comparable size.

A vision-language-action model for long-horizon and dexterous tasks showcases strong capabilities in generalization to novel objects, environments, and instructions involving abstract concepts.

An open-source multimodal agent built upon a powerful vision-language model, capable of effectively performing diverse tasks within virtual environments.