DeepMind has introduced Genie 4, a world model designed to generate interactive virtual environments in real time. Unlike traditional game engines, the system creates experiences frame by frame based on user prompts without requiring pre-rendered assets. The company views this technology as a significant step toward developing artificial general intelligence.
The new model integrates Gemini 3.8 Flash to manage simulation goals and maintain context across generated frames. While the system allows for real-time adjustments to parameters like weather or camera angles, DeepMind notes that environmental consistency is currently limited to a duration of a few minutes.
Genie 4 features improved physics simulation, specifically targeting aerodynamics and collision detection. The model also introduces native audio support, including ambient background sounds and interactive human speech that adapts to changes within the simulation. Human characters in the generated scenes are designed to exhibit more natural movement.
Google plans to release Genie 4 to developers through the Gemini API and AI Studio. For consumers, access will be provided in phases, starting with the AI Ultra subscription tier, which costs $100 per month. Future availability may extend to AI Pro and AI Plus plans, though pricing for developer access remains unannounced.
The project remains in an active research phase, with a small group of testers currently evaluating the system. Google has opened a waitlist for users interested in early access, though no specific timeline for a wider release has been established. Usage will be managed through the existing Project Genie framework.
