Google Announces Genie 4 AI World Model with Enhanced Realism and Audio Generation
Google has unveiled Genie 4, its latest generative world model designed to create interactive, simulated environments. The new model introduces significant upgrades over its predecessor, including improved visual realism, real-time environment updates via natural language prompts, and the ability to generate accompanying audio. Powered by Gemini 3.8 Flash, Genie 4 can produce experiences lasting up to two minutes. The company is positioning the technology as a versatile tool for filmmakers, game developers prototyping ideas, and robotics researchers creating training scenarios. While offering expanded capabilities, Google noted technical limitations regarding object permanence and detail consistency during longer simulations. The model will initially roll out as a research preview for trusted testers before becoming available to Google AI Ultra subscribers.