Google DeepMind's Genie 4 World Model Generates Sound and Dialogue With Its Worlds and Takes Text Edits in Real Time, Starting With Trusted Testers
Google DeepMind announced Genie 4 on October 27, a world model that generates interactive simulations together with music, ambient sound and dialogue, lets users change a running world with text prompts, and uses Gemini 3.8 Flash as an agent that monitors and steers each scene. It is available as a research preview to trusted testers, with Google AI Ultra subscribers and early-access Gemini API partners to follow. The Verge reports experiences can last about two minutes, up from one for Genie 3.
Google DeepMind's Genie 4 hands part of the job of running a simulated world to another AI model, which DeepMind says plans and checks each scene as it unfolds. Announced on October 27, the world model generates an interactive scene along with its sound. It is available only as a research preview for trusted testers.
What Genie 4 adds#
DeepMind says Genie 4 is "natively multimodal," generating music, ambient noise and interactive dialogue at the same time as the world itself, and calls it "our first real-time world model to generate audio natively." Characters can talk and react to user input or events in the scene in real time, it says.
Users can steer a running simulation with natural-language prompts, such as changing the weather, moving the camera or adding a person or object, and DeepMind says Gemini adapts the prompt to fit the scene. The model draws on the Gemini family: DeepMind says Gemini Omni anchors simulations in real-world appearance and behaviour, while Gemini 3.8 Flash acts as "a guiding agent in the loop," observing the scene, breaking goals into steps and checking incoming frames to confirm each step succeeded.
The Verge reports that Genie 4 can create experiences about two minutes long, up from one minute for Genie 3, a figure that does not appear in DeepMind's post.
DeepMind also describes experiments combining Genie 4 with Gemini 4 Argon, as part of bringing its research together with Gemini Robotics. In one example, Argon breaks a task into steps that Genie 4 renders, simulating Apptronik's Apollo humanoid robot playing checkers.
Who can use it#
DeepMind says Genie 4 has been released as a research preview to trusted testers in fields including film, journalism, education, gaming and engineering, and is "coming soon" to Google AI Ultra subscribers in Project Genie and to early-access partners of the Gemini API and Google AI Studio. A waitlist is open. It has not given pricing or a date.
All video and audio generated by Genie 4 will carry SynthID, Google's digital watermark for AI-generated content, DeepMind says, and the model was put through internal safety evaluations before the preview.
What DeepMind says still falls short#
DeepMind calls Genie 4 "an early, experimental model." It says visual details and object permanence "can still shift or drift over time," especially as a simulation runs longer, after several complex prompt edits or when moving through the scene. It lists complex physical interactions, real-time audio accuracy, legible text, spatial accuracy and long-term consistency as areas still under research.
Limits#
- The capability descriptions and demonstrations are DeepMind's own; The Verge's reporter said he had not been able to try Genie 4.
- The two-minute session length is The Verge's figure; DeepMind's post does not state a length.
- DeepMind has not published benchmark comparisons with Genie 3 or other world models in the post we read.