Google DeepMind has launched Project Genie, a web-based experience that allows users to generate and explore interactive AI-created worlds in real time, arriving months after the company previewed the Genie 3 model that powers it.

Users prompt a setting and character, preview the scene using Gemini, and then navigate an explorable world in first- or third-person. The system supports movement and exploration (walk, fly, or drive), while maintaining visual consistency by remembering what it has built so returning areas remain coherent.

Due to compute costs, sessions are currently capped at 60 seconds, with each user receiving a dedicated compute slice while exploring. Google framed this as an early step toward scalable, real-time world simulation experiences.

Access is initially limited to Google's AI Ultra tier subscribers (reported as $250/month), with plans to expand to additional tiers over time.

Why it matters: Real-time world simulators could become a foundational layer for robotics training, game development, design, and interactive media. With multiple labs pushing toward simulation-driven systems, Project Genie is a concrete step toward making “simulated reality” usable by everyday users.