We’re announcing Genie 4, our new world model for real-time interactive simulation.

Google announced Genie 4, a new world model designed to enhance real-time interactive simulations with greater realism and native audio generation. The model leverages capabilities from the Gemini family, including Gemini Omni for real-world appearance and behavior, and other Gemini models for maintaining simulation coherence. Users can initiate events and adjust scenes dynamically using natural language commands, such as altering weather or introducing new characters. The system generates ambient sounds, music, and interactive dialogue in real time, enhancing immersion and interactivity within the simulated environments.

Genie 4 represents a step toward artificial general intelligence by enabling safe training environments for robots and agents. It also supports creative industries, including filmmaking, journalism, gaming, and manufacturing, by providing tools for dynamic world-building. The model’s ability to simulate complex scenarios with native audio broadens its applications across research and commercial use cases. Google emphasizes its potential to accelerate innovation in simulation-driven technologies.

Access to Genie 4 is currently limited, with a phased rollout to a small group of trusted testers. The model will soon be available to Google AI Ultra subscribers in Project Genie and early access partners through the Gemini API and AI Studio. This approach allows Google to gather feedback while refining the model’s capabilities before broader release.

Demonstrations of Genie 4 are accessible via the Google DeepMind blog, offering a firsthand look at its features and potential applications. The announcement highlights Google’s ongoing efforts to advance AI-driven simulation technologies, positioning Genie 4 as a tool for both research and practical deployment in interactive environments.