Genie 4 adds real-time audio to Google's world model

Google DeepMind released Genie 4, its newest world model, in research preview on October 27, 2026.

The model generates interactive, photorealistic worlds, and it lets you type prompts mid-simulation to change the scene. It also generates real-time audio that matches the simulated world, including music, ambient noises, and interactive dialogue. DeepMind announced the release on its own blog, and TechCrunch's Rebecca Bellan reported the additional detail.

Genie 4 is natively multimodal, meaning the audio is generated alongside the visuals rather than added afterward. DeepMind's product page describes leaves crunching underfoot, crowds cheering, and characters who speak both with the player and with each other.

A world model generates a playable environment rather than a fixed clip, so the scene keeps running and responding to what you do. Genie is DeepMind's line of these models.

Genie 4 follows Genie 3, which Google released as a research preview a little over a year earlier, according to TechCrunch.

DeepMind said during a press briefing that the real-time behavior is the point: a world you can steer as it runs, without waiting on a render.

The team acknowledged that Genie 4 still has gaps: details and objects can drift the longer a simulation runs, and the audio's accuracy and legible in-world text are not yet solved, as TechCrunch reported.

The launch lands about two months after Runway shipped GWM Worlds 2, its second world model, which is also playable, responds to user input, and generates audio. TechCrunch notes the field is consolidating, with World Labs being acquired by AMD, Amazon in talks to buy Decart, and Rivian having acquired Third Dimension.

For creators who build scenes, Genie 4 could offer a way to prototype shots and check pacing before committing to a render. That depends on access widening past the current waitlist.

Genie 4 is open now only to trusted testers; DeepMind says it will reach Google AI Ultra subscribers in Project Genie, its tool for generating interactive worlds from text or image prompts, and early-access partners on the Gemini API and Google AI Studio in the coming weeks, and the waitlist is open now.