Google DeepMind’s Genie 4 world model adds native audio and real-time steering
Google DeepMind's Genie 4 world model adds native audio and real-time prompt steering, holding scenes for up to two minutes. It's in research preview now.
Google DeepMind announced Genie 4 on Tuesday, a new world model that generates photorealistic interactive scenes with native audio and lets users steer them mid-simulation with plain-language prompts. The model enters a research preview for a limited group of trusted testers, with a wider rollout expected in the coming weeks.
The headline addition is sound. According to Google, every world generates native audio on the fly, including ambient sounds, music and interactive dialogue. You can also set an event in motion and change it as it runs — moving the camera, switching the weather or dropping in new characters — by typing a prompt rather than scripting it.
Google says Genie 4 builds on capabilities from the Gemini family: Gemini Omni anchors simulations in real-world appearance and behavior, while other Gemini models supply the reasoning needed to keep a scene coherent. In the company’s framing, simulating the world is a step toward AGI, with safe training environments for robots and agents as one goal, alongside creative tools for filmmaking, journalism, gaming and manufacturing.
Two minutes of consistency, and a Gemini director
Independent coverage from TechCrunch adds detail Google’s post leaves out. Genie 4 can hold a consistent scene for up to two minutes, double the limit of its predecessor, Genie 3, which arrived as a research preview a little over a year ago. It also offers better physics and more human-like characters, per that report.
Gemini’s role is described as that of a director: planning steps and checking frames while Genie produces the visuals. In one experiment tied to Gemini Robotics, a Gemini 4 Argon model broke tasks into steps that Genie 4 simulated in real time — a possible route to testing robot scenarios in simulation before they happen in the real world.
“What we’ll increasingly see in the future are these hybrid world models, where you have a combination of Gemini writing code and text, and Genie ingesting that in real time and simulating reality at different levels of abstraction,” Tejas Kulkarni, research director at Google, said during a briefing.
The team is candid about the gaps. Details and objects can drift the longer a simulation runs, and TechCrunch reports the team has not fully solved keeping audio accurate or in-world text legible. Google DeepMind engineer Diego Rivas stressed during the briefing that it is still early days and that the end-to-end product experience is not yet clear.
Where you can try it
Genie 4 is available now as a research preview for trusted testers. Google says it is coming soon to Google AI Ultra subscribers in Project Genie — the company’s tool for generating interactive worlds from text or image prompts — and to early access partners in the Gemini API and AI Studio. Demos are on the Google DeepMind blog.
Related reading
Sources
- We’re announcing Genie 4, our new world model for real-time interactive simulation. — Google Blog
- Google’s Genie 4 world model adds audio, more realistic worlds, and real-time steering — TechCrunch — AI
Some links on this page may be affiliate links. If you buy through them we may earn a commission at no extra cost to you. See our affiliate disclosure.