On January 29, Google DeepMind unveiled the Project Genie app.

This prototype combines the Genie 3, Gemini, and Nano Banana Pro AI models to create interactive worlds. Currently, Project Genie is only available to US users with a $200/month Google AI Ultra subscription. The service will be rolled out to other regions later.



Here's what you can do in Project Genie:

1.   Create a world based on a description. You can specify it with a text query or images, and before generation, preview the game world and specify additional parameters. For example, specify the main character's appearance, how they will move, and whether they will be in first-or third-person view.

2.   Explore a generated world. Project Genie generates a world in real time based on user actions. Arrow keys are used to control the camera, and the WASD keys are used for movement. According to a journalist from The Verge who tested the service, worlds are generated at 720p resolution and 24 frames per second.

3.   "Mix" worlds. That is, create new interpretations of game spaces based on your own or others' prompts. The Project Genie feed displays videos of other users' generations—you can use them as a basis for your own.

Project Genie has limitations. The maximum generation time is 60 seconds. Google also notes that the image may sometimes become less realistic, and the world may not always match the user's request or the laws of physics. There may also be a delay in character control.

Users who received early access to Project Genie shared their results on social media. The service works better than any other interactive world generator.

However, The Verge journalist believes that these can't be called full-fledged games. Project Genie doesn't generate a plot, activities, or goals for the player, and it often loses consistency, "forgetting" elements of the world.

Google unveiled a preview version of Genie 3 back in August 2025. It was capable of generating interactive worlds in 720p resolution at 24 frames per second. These worlds remained consistent for several minutes.

Overall, Project Genie showcases Google DeepMind’s growing ambitions in AI-generated interactive environments. While the technology is still limited and far from delivering complete, story-driven games, it represents a significant step toward real-time world generation and creative experimentation. As Google continues to refine the system and expand access beyond early adopters, Project Genie could become an important foundation for future AI-powered game development and immersive digital experiences.

Post a Comment

Previous Post Next Post