Genie 4: DeepMind’s AI World You Can Walk Through, Change, and Hear

Google DeepMind’s Genie 4 builds a world you can walk through, change with typed text and hear, with speech and music. TechCrunch reports scenes stay consistent for up to two minutes, and only trusted testers have it now.
artificial-intelligence
Author

Kabui, Charles

Published

2026-11-02

Keywords

genie-4, google-deepmind, world-model, native-audio, real-time-simulation, robot-training

Download as PDF

Genie 4 is a Google DeepMind AI model that builds a world as you move through it, for teams that train robots or make games and films. You walk and aim the camera with the keyboard and type text to change the scene, such as the weather. A video generator returns a finished clip, but Genie 4 runs in real time: it draws each moment after your input. DeepMind calls it a world model: a chatbot predicts the next word, and a world model predicts the next moment in a place. It is also natively multimodal: one model makes the picture and the sound together, including speech and music, which earlier DeepMind world models lacked.

DeepMind names two uses: training environments for robots and AI agents, and new interactive experiences. DeepMind shows its Gemini 4 Argon model breaking a checkers game into steps while Genie 4 simulates Apptronik’s Apollo humanoid robot playing it. DeepMind says testers use it as a draft mode where a filmmaker can run several takes.

DeepMind gives no resolution or frame rate for Genie 4, while Genie 3 ran at 720p and 24 pictures a second. DeepMind says sessions last a few minutes, and TechCrunch reports scenes stay consistent for up to two minutes. Genie 4 is a research preview for trusted testers and, DeepMind says, coming soon to Google AI Ultra subscribers in Project Genie, Google’s world-building app, and early-access partners in the Gemini API and Google AI Studio. A waitlist is open.

Read More: Tencent’s HY-World 2.0, which turns text or a photo into a 3D world you can export and World Labs Atlas, which builds 3D scenes and robot training simulations

Sources:

Disclaimer: For information only. Accuracy or completeness not guaranteed. Illegal use prohibited. Not professional advice or solicitation. Read more: /terms-of-service

Reuse

GNU GENERAL PUBLIC LICENSE v3.0(View License)

Citation

BibTeX citation:
@misc{kabui2026,
  author = {{Kabui, Charles}},
  title = {Genie 4: {DeepMind’s} {AI} {World} {You} {Can} {Walk}
    {Through,} {Change,} and {Hear}},
  date = {2026-11-02},
  url = {https://toknow.ai/posts/genie-4-deepmind-real-time-world-model-native-audio/},
  langid = {en-GB}
}
For attribution, please cite this work as:
Kabui, Charles. 2026. “Genie 4: DeepMind’s AI World You Can Walk Through, Change, and Hear.” https://toknow.ai/posts/genie-4-deepmind-real-time-world-model-native-audio/.