Google says Genie 4 creates more realistic interactive simulations

Google DeepMind announced Genie 4 as an update to its world model for creating real-time interactive simulations. The company says the model generates more realistic worlds, native audio—including ambient sounds and interactive dialogue—and lets users change the weather, camera, people, or objects with natural-language prompts during a simulation. Gemini 3.8 Flash acts as a guiding agent, reviewing frames and helping maintain scene coherence. Google says Genie 4 improves the simulation of physics and human movement and can keep a scene consistent for up to two minutes, twice Genie 3’s limit.

The model has begun rolling out as a research preview to trusted testers. In the coming weeks, it will be offered to Google AI Ultra subscribers in Project Genie and to early access partners of the Gemini API and Google AI Studio. The company acknowledges that details and objects can drift over time, and that it still needs to address problems with audio accuracy and the legibility of text inside simulations.