Project Genie Community Gallery
In brief
Project Genie introduces a community gallery alongside Genie 4, letting eligible subscribers generate and edit interactive 3D worlds in real time with text and audio 1. You can also build worlds using conversational agents and publish them for others to explore 1.
Who can use it
The experience is available today for eligible Google AI Ultra subscribers 1. Google AI Pro members can join a waitlist on the same page 1.
Does this affect you?
- DevelopersYou can explore frontier world models like Genie 3 and Genie 4, though direct API access details are not provided yet.
- Product managersYou can study how multimodal AI agents and community remixing features drive user engagement and interactive content creation.
- FoundersYou can evaluate new paradigms in generative entertainment and interactive media for potential product opportunities.
- Everyday usersYou can explore immersive community worlds, generate your own environments using plain text, and share them easily.
How to start
Official step-by-step setup instructions are not published yet, but you can access the platform directly through the Project Genie website 1.
Visit the Project Genie Website
Go to the official Project Genie website to start building your own worlds if you are an eligible subscriber 1.
Google Blog
What you can build
- Real-Time World EditingType prompts mid-session to change the weather, add characters, or alter the environment while navigating your generated world.
- Community World RemixingBrowse curated and community-built spaces in the gallery, then remix them to create your own unique interactive experiences.
Before you rely on it
Quick check
3 questionsAvailability
Sources
2 merged into this cardMore from Google

Gemini 4 Neon
Google's new Gemini 4 model pairs a 1M-token context window and four thinking levels with lower cost per task. It scores 61 on the AA Intelligence Index and targets complex coding, analysis, and cyber workflows.
Google updates Genie 3 model
Google has updated the Genie 3 world model with 3-minute sessions, world chaining, 360-degree panorama grounding, and real-time promptable events.