Google DeepMind Announces Genie 4 Interactive World Model

Google DeepMind has announced research on Genie 4, a new world model built for real-time interactive simulation. According to the company, the release brings realism and native audio generation to real-time interactive simulated environments.

According to Google DeepMind, Genie 4 incorporates and builds on capabilities from the Gemini model family. Within this setup, Gemini Omni anchors simulations in real-world appearance and behavior, while other Gemini models provide the complex reasoning needed to help maintain a simulation's coherence and continuity.

With Genie 4, users can set an event in motion and steer scenes in real time using natural language. According to Google DeepMind, these inputs allow users to make changes such as moving the camera, changing the weather, or introducing new characters.

Every world produced by Genie 4 also generates native audio on the fly, including ambient sounds, music, and interactive dialogue.

Google DeepMind stated that simulating the world is a step toward AGI, enabling safe training environments for robots and agents. The company also noted that the model can support creative tools across industries such as filmmaking, journalism, gaming, and manufacturing.

Google DeepMind has started rolling out Genie 4 to a limited group of trusted testers. The company plans to make the model available to Google AI Ultra subscribers in Project Genie, as well as early access partners in the Gemini API and AI Studio. Demos of the model have been published on the Google DeepMind blog.