Genie 3
Google·
Overview
Google DeepMind's first real-time interactive general-purpose world model. From a text prompt it generates a navigable environment at 24 frames per second and 720p that stays consistent for a few minutes, and the world can be changed mid-simulation with further text prompts, which DeepMind calls promptable world events. Released as a limited research preview and positioned as a training environment for embodied agents rather than as a video generator; a public demo followed in January 2026 as Project Genie.
Capabilities and innovations
Benchmarks
No established, comparable metric exists for world models yet. A video model goes from prompt to clip; a world model goes from an image or scene to a navigable space with camera control, which no video or image board measures. No scores and no crown until one exists.
Links
More from Google
Data curated by AI Model Timeline. See the methodology for admission criteria, benchmark eras and source priority.