Google DeepMind releases Genie 4, a real-time simulator of interactive worlds
The new world model generates photorealistic scenes that can be explored with a keyboard and steered with text as they run, with native sound and dialogue. It is available as a research preview to trusted testers.
Source published · In the Daily Brief of 28 October 2026
Filed under AI industry, Agents
Google DeepMind announced Genie 4 on 27 October. A user navigates a generated world with standard keyboard and camera controls and can change it mid-simulation in plain language, for example altering the weather, moving the camera or adding a person or object. Between inputs, a Gemini model monitors the scene and steers it so that the world keeps evolving coherently.
The model is natively multimodal: ambient sound, music and spoken dialogue are generated together with the images. DeepMind says Gemini Omni anchors the appearance and behaviour of scenes in the real world, while other Gemini models supply the reasoning that keeps a simulation consistent over time. The company highlights more realistic physics, such as collisions and aerodynamics, and its largest gain in the simulation of people, who move naturally and can be steered in real time.
Genie 4 is rolling out as a research preview to trusted testers, with later access planned for Google AI Ultra subscribers in Project Genie and early partners of the Gemini API and Google AI Studio. DeepMind presents world models as a source of varied, safe training environments for robots and agents as well as a medium for interactive experiences. The announcement was the most discussed AI story of the day among researchers.