RUNWAY RELEASES ITS FIRST WORLD MODEL, ADDS NATIVE AUDIO TO LATEST VIDEO MODEL

Summary from the TLDR Design newsletter

Runway's GWM-1 is a world model that simulates physics and real-world behavior through frame-by-frame prediction. There are specialized versions for interactive environments (GWM-Worlds), robotics training (GWM-Robotics), and human avatars (GWM-Avatars). Runway positions GWM-1 as more general-purpose than competitors like Google's Genie-3, targeting applications in robotics, life sciences, and agent training. The company has also updated its Gen 4.5 video model to include native audio capabilities and long-form multi-shot generation, enabling users to create one-minute videos with consistent character and dialogue.