Google DeepMind announces Genie 4, a real-time world model with native audio

What is changing?

Google DeepMind announced Genie 4 on October 27. The new world model generates photorealistic visuals together with ambient sound, music and interactive dialogue in real time. Gemini Omni grounds simulations in real-world appearance and behavior, while Gemini 3.8 Flash operates in the loop as a guiding agent that continuously observes scenes and tracks multi-step goals.

Users can change weather, camera angles, people or objects with natural-language instructions while a simulation is running. DeepMind also reports progress in physical events such as aerodynamics and collisions, natural human motion, context-sensitive dialogue and multi-step events planned with Gemini models. The official launch video demonstrates these real-time interactions.

Genie 4 is currently an experimental research preview for trusted testers. It is planned for later access through Project Genie for Google AI Ultra subscribers and early-access partners of the Gemini API and Google AI Studio. Detail and object persistence over long sessions, complex physics, audio accuracy, legible text and spatial consistency remain limited, and generated video and audio will carry SynthID.