Google DeepMind releases Genie 4 world model with real-time audio and mid-simulation prompts

Google DeepMind on Tuesday 27 October released Genie 4, the latest version of its “world model”, an AI system that generates interactive virtual environments that users can move through and change as they go.

Google said users can steer a scene in real time with natural-language prompts, for example moving the camera, changing the weather or adding new characters, and that each world now generates its own audio, including ambient sound, music and dialogue. The company said the model draws on its Gemini family of models, with Gemini Omni used to keep simulations close to real-world appearance and behaviour.

According to TechCrunch, Genie 4 can keep a scene consistent for up to two minutes, double the limit of Genie 3, which Google released as a research preview a little over a year ago. Google researchers told reporters at a briefing that details can still drift during longer simulations and that audio accuracy and on-screen text remain unsolved problems, TechCrunch reported.

Google said simulating the world is “a crucial step toward AGI”, and listed training environments for robots and AI agents and creative tools for filmmaking, gaming and manufacturing among possible uses.

The model is available now to a limited group of trusted testers. Google said it will come to Google AI Ultra subscribers through its Project Genie tool, and to early-access partners of the Gemini API and AI Studio; TechCrunch reported that rollout is planned for the coming weeks.