Oct 28, 2026

Genie 4: Google DeepMind's real-time world model now generates sound

Genie 4 Google DeepMind Gemini Project Genie World models

Genie 4 is Google DeepMind's new real-time world model: it generates playable worlds with native audio, takes text steering through Gemini, and adds more realistic physics and people. What it does, who gets it, and the catch.

Sir Toastimir

Footage: Google DeepMind

Genie 4 is Google DeepMind's new generation of world simulation, announced Oct 27, 2026. It generates interactive worlds in real time, and DeepMind calls it its first real-time world model to generate audio natively.

What's new

Genie 4 is natively multimodal: music, ambient noises and interactive dialogue are generated at the same time as the world. You move with standard keyboard controls for navigation and camera, and you can steer the world with natural language, like changing the weather, moving the camera, or introducing a new person or object. DeepMind says Gemini adapts your prompt and helps monitor and steer the world between inputs.

More realistic worlds

DeepMind highlights more photorealistic visuals, more realistic physics such as collisions and aerodynamics, and more natural people who can converse and react in real time. TechCrunch reports that worlds stay consistent for up to about two minutes, double Genie 3, and DeepMind's model page describes a few minutes of continuous interaction.

What people are using it for

DeepMind says trusted testers have used Genie 4 for film pre-visualization, bringing archival photos to life, prototyping games and simulating enterprise scenarios. It also showed Gemini 4 Argon and Genie 4 working together to simulate Apptronik's Apollo humanoid playing a game of checkers.

The catch

DeepMind notes that details and object permanence can still drift over time. All video and audio generated by Genie 4 includes a SynthID watermark.

Who gets it

Genie 4 is available to trusted testers now. Google says it is coming soon to Google AI Ultra subscribers in Project Genie and to early-access developers through the Gemini API and Google AI Studio.

Visuals in the video

All footage is from Google DeepMind's Genie 4 announcement and model page.