Google DeepMind has introduced Genie 4, its latest AI world model capable of generating photorealistic, interactive environments with real-time audio and natural language controls. Released on October 27, 2026, the model allows users to modify simulated scenes as they unfold, with potential applications in gaming, filmmaking, education and robotics.
According to Google DeepMind’s official announcement, Genie 4 improves on its predecessor with more realistic physics, humanlike characters and the ability to generate synchronized music, environmental sounds and interactive dialogue. The model can maintain a consistent simulated environment for up to two minutes, approximately twice the duration supported by Genie 3.
Genie 4 Adds Real-Time Editing and Gemini Integration
Users can modify virtual environments through text prompts, including changing weather conditions, adjusting camera angles and introducing new characters or objects. The model responds dynamically to these changes while simulating physical interactions such as collisions, movement and environmental effects.
Genie 4 also integrates more closely with Google’s Gemini models, which help interpret instructions, plan sequences of actions and monitor generated scenes. Gemini 3.8 Flash can guide simulations by breaking tasks into steps and checking their progress as Genie 4 generates the environment.
Google DeepMind has also tested Genie 4 alongside Gemini 4 Argon in robotics research, allowing AI systems to simulate multi-step tasks before attempting them in physical environments. The research examined how Gemini could direct simulations and adjust instructions as conditions changed.
Filmmaking and Interactive Learning Among Potential Applications
Early testers have identified film production as a potential application, particularly during concept development and storyboarding. Genie 4 allows creators to test scenes, modify camera movements and experiment with dialogue without producing separate video sequences for each change.
Google is also examining applications in interactive education, including simulated historical environments where users could interact with virtual characters. The company already offers Veo for AI video generation, while its Gemini Omni model could eventually help convert interactive prototypes into finished video content.
Genie 4 arrives amid growing competition in AI-powered world simulation, following Runway’s GWM Worlds 2 release and several acquisitions involving simulation technology. These include AMD’s planned $8.2 billion acquisition of World Labs and Rivian’s acquisition of Third Dimension for autonomous driving research.
Research Preview Expands to Google AI Ultra Subscribers
Google acknowledged that Genie 4 remains experimental, with limitations involving visual consistency, object permanence, audio accuracy and text rendering. Simulated details can change unexpectedly during longer sessions, particularly after repeated modifications.
The model is currently available to selected trusted testers, with access expanding to Google AI Ultra subscribers through Project Genie in the coming weeks. Early access partners using the Gemini API and Google AI Studio will also receive access, although Google has not announced a broader public release date.
Featured image credits: Wikimedia Commons
For more stories like it, click the +Follow button at the top of this page to follow us.