Models · 2026-10-27
Google unveils Genie 4: AI that creates interactive worlds with sound and dialogue
Google DeepMind has announced Genie 4, a new model that creates interactive virtual worlds in real time, with realistic visuals, sound, and character dialogue.
What Genie 4 is
Google DeepMind has introduced Genie 4, a new "world model". This is AI that can create a virtual environment you can move through, and which reacts to actions. In simple terms, you describe what should happen, and Genie 4 shows how the scene unfolds.
According to the developers, Genie 4 generates both visuals and sound at the same time: wind, waves, music, and even people talking in the scene. Earlier models of this kind mostly produced only images.
How it works
Users control the camera with standard keyboard keys and can change the scene with text prompts. For example, they can change the weather, move the camera, or add a new person or object. The model adapts the prompt to the surrounding scene.
Between user inputs, Gemini, another Google model, monitors the scene and keeps the world consistent. Gemini Omni anchors the visuals in real-world appearance, while other Gemini models help maintain coherence and continuity.
What makes Genie 4 different
Google highlights three things: more photorealistic visuals, natural human movement, and physics such as collisions and aerodynamics. Characters move and react to their environment and can be steered in real time.
The company also emphasizes speech. Characters can talk and react to events in the world or to user actions during the simulation.
Who can try it
For now, Genie 4 is a research preview for trusted testers. Wider access will start with Google AI Ultra subscribers in Project Genie, along with early access partners of the Gemini API and Google AI Studio.
To get on the waitlist, people need to go to Project Genie. Developers say they will gather feedback first and only then release the model to consumers and developers more broadly.
Why it matters
Google sees such models as an important step toward general-purpose AI, meaning systems that understand the world the way people do. Virtual worlds can also be used to train robots and AI agents and to create new interactive experiences.
For an ordinary user this is still a future development, but the direction of games and virtual worlds is already shifting.
Sources: 🔗 Google DeepMind Blog