Google DeepMind Releases Genie 4 World Model With Audio and Live Steering

Google DeepMind released Genie 4, its newest world model, as a research preview for trusted testers, TechCrunch reported. It will roll out to Google AI Ultra subscribers in Project Genie and to early-access partners of the Gemini API and Google AI Studio in the coming weeks.

Genie 4 generates more photorealistic interactive worlds than Genie 3 and adds real-time audio, including music, ambient sound and dialogue. Users can type prompts mid-simulation to change the weather, camera angle or add people and objects. It keeps a scene consistent for up to two minutes, double its predecessor, with better physics.

Gemini works with Genie 4 as a director that plans steps and checks frames. In a Gemini Robotics experiment, a Gemini model broke tasks into steps that Genie 4 simulated in real time. Google said details can still drift during long simulations and in-world text is not always legible.