Google DeepMind has announced Genie 4, a real-time interactive world model that generates explorable environments with native audio.
Genie 4 builds a world as it is explored, rendering, in DeepMind's wording, "how the world looks, sounds, and responds, pixel by pixel." Users move through a scene with standard keyboard controls and steer it with text prompts, which Gemini adapts to match the surrounding scene. DeepMind describes no mesh, splat or other 3D output from the model. World Labs' Marble exports its worlds as Gaussian splats and meshes, and SpAItial's Echo-3 builds real-scale 3D scenes from iPhone captures.
Genie 4 draws on two named Gemini models. Gemini Omni anchors the simulations in real-world appearance and behavior. Gemini 3.8 Flash serves as a guiding agent during real-time generation, monitoring and steering the scene between user inputs and adapting its plan as the world changes. DeepMind is also experimenting with pairing Genie 4 and Gemini 4 Argon.
Genie 4 is DeepMind's first real-time world model to generate audio natively, producing music, ambient sound and interactive dialogue alongside the visuals. Characters converse and react to user inputs or events in the world. All video and audio generated by Genie 4 will carry SynthID watermarking.
DeepMind says Genie 4 can maintain coherent environments over short sessions, but visual details and object permanence "can still shift or drift over time," with drift more likely in longer simulations, after multiple complex prompt edits, or while navigating. Complex physical interactions, real-time audio accuracy, legible text, spatial accuracy and long-term consistency remain active research areas. DeepMind does not publish resolution, frame rate or session length figures.
Genie 4 is coming soon to Google AI Ultra subscribers through Project Genie, and to early access partners in the Gemini API and Google AI Studio. Genie 4 is available as a research preview to trusted testers, with broader access to follow.
