Agentic WeeklyGet the weekly

Google DeepMind announces Genie 4, a real-time world model with native audio, for trusted testers first

Release Ruler for this brief: marked at its publication date, Oct 27, 2026, among the AI models we track.

Genie 4 is Google DeepMind's new world model for real-time interactive simulation, announced Oct 27, 2026. Scenes can be steered with natural language and generate native audio; it is rolling out to trusted testers first, with Google AI Ultra and Gemini API early access listed as coming soon.

By Agentic Weekly · Evidence: Reported · How we work

First seen
(via Google DeepMind blog (blog.google))
Source published
(we saw it 19 minutes later)
Brief published

What changed

  • Google DeepMind announced its research on Genie 4, which it describes as a world model for real-time interactive simulation with greater realism and native audio.
  • The post says Genie 4 builds on the Gemini family: Gemini Omni anchors simulations in real-world appearance and behavior, and other Gemini models supply reasoning to keep a simulation coherent.
  • Users can set an event in motion and steer a scene in real time with natural language, for example moving the camera, changing the weather or adding characters.
  • Every world generates native audio on the fly, including ambient sound, music and interactive dialogue, according to the announcement.
  • Availability: rolling out now to a limited group of trusted testers; the post 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.

In the source’s words

“We have started rolling out Genie 4 to a limited group of trusted testers.”

“Simulating the world is a crucial step toward AGI, enabling safe training environments for robots and agents”

What we have not verified

  • We did not run Genie 4; it is limited to trusted testers and we have no access.
  • The announcement gives no dates, pricing, rate limits or API model name for the Gemini API, AI Studio or Project Genie rollout.
  • No benchmarks, resolution, frame rate or session length figures are stated in the post, so we report none.
  • We did not review the demos on the Google DeepMind blog.

This brief is Reported: written from the sources below, not from running the release ourselves.

Sources

Spotted an error? See our corrections policy or email the address on the contact page. Markdown version for agents: /news/google-deepmind-genie-4-world-model.md

Want these briefs in your Google Top Stories? Add Agentic Weekly as a preferred source on Google.