Genie 4: Google's world model adds native audio and live steering

Google has announced Genie 4, a world model for real-time interactive simulation. It is described as research from Google DeepMind. Google says it can help create safe training environments for robots and agents, and support creative work in filmmaking, journalism, gaming and manufacturing.

Genie 4 builds on the Gemini family. Google says Gemini Omni anchors simulations in real-world appearance and behavior, while other Gemini models handle the reasoning that keeps a simulation coherent. Users can steer a scene with natural language, such as moving the camera, changing the weather or adding characters.

Every world also generates audio on the fly, including ambient sounds, music and interactive dialogue. Google says the model brings unprecedented realism. Bloomberg notes the technology has rattled investors in video game companies.

Access is limited for now. Google has started rolling out Genie 4 to a limited group of trusted testers. It is coming soon to Google AI Ultra subscribers in Project Genie, and to early access partners in the Gemini API and AI Studio.