The new AI system creates explorable virtual spaces where users can introduce objects, modify weather, or alter camera angles using text commands. Built on Gemini Omni, Genie 4 simulates real world physics, such as aerodynamics and collisions, to generate interactive environments from a single prompt. The model includes native audio for ambient sound and dialogue, and uses Gemini reasoning to plan multi-step events to ensure the scenes remain coherent as they evolve.

Google DeepMind is currently rolling out the model as a research preview to trusted testers, with future access planned for early access partners via the Gemini API and Google AI Studio. The model will also be available to Google AI Ultra subscribers through Project Genie. DeepMind is using the tool in robotics research, pairing Genie 4 with Gemini 4 Argon to test how agents navigate simulated scenarios before they are deployed in the physical world.