# Google DeepMind has introduced a model for creating interactive 3D worlds

Google DeepMind has introduced Genie 4, a model that creates interactive virtual worlds in real time. Users navigate the generated environment using a keyboard, while the model dynamically builds the surrounding space, including graphics, physics, sound, and dialogue.

Genie 4 delivers more photorealistic visuals and physics than previous versions, such as object collisions and aerodynamics. People within the world move naturally and can be controlled. The scene can be modified via text commands during the simulation: changing the weather, moving the camera, or adding a person or object.

At Google DeepMind, Genie 4 is described as the first world model that generates sound in real time, including background noise and music. Characters can respond to users and react to events in the world. To maintain scene coherence, the model works in conjunction with Gemini, which plans steps and verifies results.

Developers envision several use cases for Genie 4. Directors can quickly sketch scenes for future films, game developers can test mechanics, and engineers can simulate situations for robot training. The model can also transform archival photographs into volumetric environments. All generated video and audio are marked with an invisible SynthID watermark.

Genie 4 is currently available in closed testing. It will first be accessible to subscribers of the Google AI Ultra plan in the Project Genie service and partners with early access to the Gemini API and Google AI Studio. Interested parties can join the waitlist.

Source: [Google DeepMind](<https://deepmind.google/blog/genie-4-a-new-generation-of-world-simulation/>)
