Evergreen Tools
|
← Back to AI Tools

Genie 4: A Real-Time Interactive World Model With Native Audio

Google DeepMind announced Genie 4 on 27 October 2026: type a prompt, walk into the world with a keyboard, and change the weather mid-scene with a sentence. It is natively multimodal, generating music, ambient sound and interactive dialogue alongside the visuals, and a scene holds together for up to two minutes — double Genie 3. For now it is a research preview for trusted testers only

Tool Interface

Interactive tool will be available soon

Features

  • ✓ Natively multimodal: visuals, music, ambient noise and interactive dialogue are generated together, with no separate audio or voice model bolted on
  • ✓ Real-time steering: natural language changes the weather, moves the camera or introduces a new person or object, and Gemini rewrites your prompt to fit the surrounding scene
  • ✓ Better physics and people: aerodynamics and collisions behave plausibly, and characters move naturally and can be guided live
  • ✓ Twice the continuity: reporting says a single scene can stay coherent for up to two minutes, double the Genie 3 limit
  • ✓ Built on the Gemini family: Gemini Omni anchors real-world appearance and behaviour while other Gemini models supply the reasoning that keeps a simulation coherent

How to Use

  1. Check availability first: it is a research preview for trusted testers, so general users cannot get in yet
  2. Join the queue: Project Genie has a public waitlist, and Google AI Ultra subscribers get access next
  3. Watch for early access partner slots in the Gemini API and Google AI Studio
  4. Pitch it where DeepMind points: robot and agent training environments, film and game concept work, education and journalism scene reconstruction

FAQ

When did Genie 4 launch, and can I use it now?

It was announced on 27 October 2026; the source is https://deepmind.google/blog/genie-4-a-new-generation-of-world-simulation . It is a research preview open to a small group of trusted testers, so regular users and developers have to wait.

What is the biggest difference from Genie 3?

Two things: native audio, with music, ambient noise and interactive dialogue generated in real time with the world, and control, since you can change weather, move the camera or add people and objects mid-simulation. Reporting puts scene continuity at up to two minutes, double Genie 3.

When can I try it myself?

DeepMind says it rolls out to Google AI Ultra subscribers in Project Genie and to early access partners in the Gemini API and Google AI Studio. The Project Genie waitlist is public. No dates were given.

Is the audio added in post?

No. DeepMind states every sound in the demos is produced directly by Genie 4, including ambient sound, music and interactive dialogue, rather than pulled from an external effects library.

Can I build production on it?

DeepMind calls it an early, experimental research milestone and is still collecting feedback from testers in creative industries, journalism, education, gaming and engineering. It fits prototyping and training-environment exploration, not production dependencies.