Genie 4: A new generation of world simulation

Google DeepMind has introduced Genie 4, a research model capable of generating real-time, interactive 3D simulations that include integrated audio and conversational characters. By utilizing Gemini models for background planning, the system maintains greater structural consistency during user interactions. While currently limited to a preview for trusted testers, the technology represents a step toward high-fidelity world generation. The project highlights a shift in generative AI toward dynamic, responsive environments, though the model still struggles with visual stability during longer sessions.