Google DeepMind has announced Genie 4, a world model that generates interactive simulations in real time, including sound and dialogue. The company has started rolling it out as a research preview for trusted testers.
Users can navigate a simulated world with standard keyboard controls and steer it with text prompts. Google DeepMind says they can change the weather, move the camera or introduce a person or object while a scene is running. Gemini adapts those requests to the scene and monitors it between inputs to help keep events coherent.
Genie 4 generates sound with the scene
Genie 4 is Google DeepMind’s first real-time world model to generate audio natively, according to the company. That includes ambient sounds, music-like audio and interactive dialogue. Characters can speak and react to user input or events as they happen.
The company also says Genie 4 improves the appearance and movement of its simulations. It describes support for physical interactions such as aerodynamics and collisions, and says people in the scenes can be steered in real time.
Gemini helps guide longer sequences
Google DeepMind says Genie 4 draws on the Gemini family of models. Gemini Omni helps anchor scenes in real-world appearance and behavior. Gemini 3.8 Flash acts as a guiding agent: it observes incoming frames, plans steps toward a goal and checks their progress as the world changes.
The company says these capabilities could support training environments for robots and AI agents, as well as interactive experiences. Those are potential uses, rather than products announced for general availability.
Google DeepMind is gathering feedback from trusted testers before a wider release. It says access for consumers and developers will begin with Google AI Ultra subscribers through Project Genie and early access partners of the Gemini API and Google AI Studio. A Project Genie waitlist is open.
