Google DeepMind releases Genie 4 with real-time audio and steering

Google DeepMind has released Genie 4 as a research preview for trusted testers. The world model generates interactive scenes with real-time audio, including music, ambient sounds and dialogue, and lets users change a simulation with text prompts while it runs.

The company says Genie 4 produces more photorealistic worlds than its predecessor, Genie 3, and supports freeform edits such as changing the weather or camera angle or adding people and objects. It can maintain a consistent scene for up to two minutes, twice Genie 3’s limit, and includes improved physics and more human-like characters.

Google says Gemini can plan steps and check frames while Genie 4 generates visuals. In an experiment tied to Gemini Robotics, a Gemini 4 Argon model broke tasks into steps that Genie 4 simulated in real time. The company described this as a possible way to test robot scenarios in simulation before they occur in the real world.

Google DeepMind engineer Diego Rivas said trusted testers have identified film prototyping as a potential use, particularly during ideation and storyboarding. He said turning prototypes into creative assets with Gemini Omni remains some way off. The team also said details and objects can drift during longer simulations, and it has not yet solved how to ensure accurate audio or legible in-world text.

Genie 4 is available now to trusted testers. Google says it will reach Google AI Ultra subscribers in Project Genie and early access partners of the Gemini API and Google AI Studio in the coming weeks.