Google DeepMind announces Genie 4, a world model you steer in real time

On October 27, 2026, Google DeepMind announced research on Genie 4, a world model for real-time interactive simulation with native audio.

With Genie 4, you steer a scene in real time using natural language. Google's examples are moving the camera, changing the weather or introducing new characters. Every world also generates native audio on the fly, including ambient sounds, music and interactive dialogue.

For now, access is limited. Genie 4 is rolling out to a limited group of trusted testers. Google says it is coming soon to Google AI Ultra subscribers in Project Genie, and to early access partners in the Gemini API and AI Studio.

What that means for you depends on where you stand. If you are not one of the trusted testers, you cannot use Genie 4 yet. If you pay for Google AI Ultra, Project Genie is where it is coming. If you build with the Gemini API or AI Studio, the announced access there is for early access partners. Google describes both of these as coming soon.

Our take:

it is a research release, so watch the demos, but do not plan a project around it yet.