✳ the wire · analysis
Google DeepMind announces Genie 4, a real-time world model with native audio, in research preview
An email the minute a new frontier model answers on a public API · free, no card
Google DeepMind announced Genie 4 on October 27: a world model that generates an interactive scene in real time and lets you steer it while it runs. It is rolling out as a research preview for trusted testers, with a waitlist open in Project Genie.
WHAT DEEPMIND SAYS IT DOES
Genie 4 is natively multimodal, per DeepMind, so it generates sound along with the world: music, ambient noise and dialogue. Users move through a scene with keyboard controls and change it with text prompts as it runs, for example changing the weather, moving the camera or adding a person or object. DeepMind says it builds on the Gemini family, with Gemini Omni anchoring how scenes look and behave and other Gemini models keeping the simulation coherent.
WHO GETS IT
Trusted testers first. DeepMind says wider access will start with Google AI Ultra subscribers in Project Genie and early-access partners of the Gemini API and Google AI Studio.
WHAT WE DO NOT KNOW
A release date, pricing, output resolution, or how long a session can run. Everything above is DeepMind's own description; we have not used it.
Source: Google DeepMind's Genie 4 announcement (read directly) · @Mr_Salio, @LuminaBench and @scaling01 on X ↗ · the bench index
