Google’s Genie 4 lets you edit AI generated worlds in real time, complete with sound and dialogue

Genie 4 enables users to create and modify virtual worlds as they run, allowing for changes like new weather or objects without restarting the simulation. The model generates native audio, including dialogue, which Google suggests could be as significant for interactive worlds as it was for AI video. This iteration builds on Genie 3 with enhanced realism, more natural human movement, and improved physics simulations, including better handling of aerodynamics and collisions.

The technology could support various professional applications, from game prototyping and filmmaking previsualization to developing training environments for robots and other AI agents. Google is initially rolling out Genie 4 as a research preview to select testers, including filmmakers, game developers, and robotics teams. It also intends to make the model available to developers through the Gemini API and AI Studio, though specific pricing for these channels is not yet confirmed.

For consumers, Google has outlined a phased access plan for Genie 4. It will first be available through Project Genie to existing AI Ultra subscribers, who currently pay $100 per month. Google indicates that access for AI Pro and AI Plus subscribers may follow at a later, unspecified date. A waitlist is open for early access, but a firm timeline for broader consumer availability has not been provided. Google notes these simulations can sometimes drift after a couple of minutes, potentially losing track of details.