Google’s Project Genie Opens Access to 3D AI Worlds

This summer, Google DeepMind introduced Genie 3, a novel AI system capable of generating dynamic, interactive 3D worlds that respond as users navigate through them. Originally developed as a tool to train AI agents, Genie 3 is now being made accessible to the public through Project Genie.

Project Genie offers three modes: World Sketching, exploration, and remixing. The experience starts with World Sketching, where Google’s Nano Banana Pro model creates a source image that Genie 3 will use to build the immersive world. Users can customize their character, select a camera perspective first-person, third-person, or isometric and specify how they want to explore the environment. Before entering the interactive world, Nano Banana Pro provides a preliminary “sketch” for preview and adjustment. Users can also write their own prompts to generate worlds that others have created with Genie.

Access to Project Genie requires a subscription to Google’s AI Ultra plan, which costs $250 per month, and is currently limited to adults aged 18 or older residing in the United States. While Genie 3 generates worlds with game-like visuals and can simulate physical interactions, it does not function as a traditional game engine. There are no conventional game mechanics involved. Additionally, each generation is limited to 60 seconds, with a maximum presentation frame rate of 24 frames per second and a resolution capped at 720p.

For AI Ultra subscribers, Project Genie provides a unique chance to experience the forefront of DeepMind’s recent AI advancements in interactive world generation.