Genie 4 is the world model Google DeepMind introduced on 27 October 2026, in a post on the Google DeepMind blog published that day. It does more than generate images: it also generates ambient sound, music and character dialogue at the same time, while users move through the simulation with a keyboard and rewrite the scene with text as it runs. For games, film and robotics training, the appeal is rehearsing in a virtual world first, rather than paying the price of trial and error in the real one.
Sound is the biggest change
Genie 4 is DeepMind's first real-time world model to generate audio natively. Every world carries its own ambient sound, with wind in the trees and waves at the beach, while character dialogue changes in real time with the user's input and events in the world. That makes a simulation feel closer to a live scene that can listen, respond and be interrupted.
Scenes can be changed while they run
Users can change the weather, move the camera or add a new person or object in natural language while a simulation is running, and Gemini adapts the prompt to the surrounding scene. In the loop, Gemini 3.8 Flash acts as a guiding agent that keeps observing the scene, breaks a goal into steps and checks progress frame by frame; when the user intervenes or the world changes, it adjusts the plan on the spot. Gemini Omni anchors the simulation in the appearance and behaviour of the real world, while other Gemini models help it stay coherent.
Trusted testers first, and the limits are stated openly
Genie 4 is currently available as a research preview for trusted testers. It will then reach Google AI Ultra subscribers in Project Genie first, along with early access partners of the Gemini API and Google AI Studio. DeepMind also acknowledges that the longer a simulation runs and the more complex the edits become, the more visual details and the persistence of objects can drift, and that complex physical interaction, real-time audio accuracy, legible text rendering and spatial accuracy remain areas of research. All video and audio generated by Genie 4 will carry a SynthID watermark, so AI-generated content can be identified.
For most readers, the value of Genie 4 is not another more realistic demo. It turns world simulation into a way of rehearsing work again and again: film-makers can step into a scene and try shots, game teams can test ideas before writing engine code, and robotics teams can test workflows in a virtual environment. Until it opens to more people, though, it reads more like a research preview than a product ready to be delivered today.
