Google DeepMind has announced Genie 4, an artificial-intelligence model that creates interactive simulated worlds with visuals, sound and dialogue. Users can steer those scenes with natural-language prompts while they run, including requests to move the camera, change the weather or add characters. Genie 4 is currently available only as a research preview for trusted testers.
Genie 4 adds native audio
Unlike a standard video generator, Genie 4 is built for worlds that respond as users explore them. Google says each world can produce ambient sounds, music and interactive conversations as the scene develops. The model builds on the Gemini family to improve the appearance, behavior and continuity of its simulations. (Google announcement)
Robotics and creative testing
Google says these interactive worlds could help researchers test AI agents and robots in simulated environments before they operate in the physical world. The company also points to possible uses in filmmaking, journalism, gaming, manufacturing and interactive learning. In one research example, Gemini creates a multistep plan while Genie 4 simulates the results in real time. (Google DeepMind details)
Access remains limited
For now, Genie 4 is available to only a limited group of trusted testers. Google says it will next come to Google AI Ultra subscribers through Project Genie and to early-access partners using the Gemini API and Google AI Studio. The announcement does not give a public release date or price.
Google lists major limits
Google describes Genie 4 as an experimental system. DeepMind says visual details and objects can drift during longer sessions or after complex edits. Real-world locations may not be reproduced accurately, text may be difficult to read, audio can be inaccurate and continuous interaction is limited to a few minutes. Video and audio generated by the model will include Google’s SynthID watermark. (Genie 4 limitations and safety notes)
References
Google. (2026, October 27). We’re announcing Genie 4, our new world model for real-time interactive simulation. Google Blog. https://blog.google/innovation-and-ai/models-and-research/google-deepmind/introducing-genie-4/
Google DeepMind. (n.d.). Genie 4. Retrieved October 28, 2026, from https://deepmind.google/models/genie/
Bellan, R. (2026, October 27). Google’s Genie 4 world model adds audio, more realistic worlds, and real-time steering. TechCrunch. https://techcrunch.com/2026/10/27/googles-genie-4-world-model-adds-audio-more-realistic-worlds-and-real-time-steering/