Imagine a neighborhood square at noon. You cross the street using the keyboard and ask for rain. That’s the kind of experiment Google is pitching for Genie 4, which reached the Gemini API and Google AI Studio on November 19. The square is our example, not a scene we tested.
What changes as you explore
Start with a description, a reference image, or both. As you move, you can ask for different lighting, weather or objects. The model card describes real-time 720p video at up to 24 frames per second with synchronized audio. The browser-based Project Genie is a separate option that Google announced for eligible Google AI Ultra subscribers on November 18.

Why it isn’t a game editor
The interesting use is finding out quickly whether an idea deserves a level. Does the same square invite exploration in daylight and feel suspenseful at night? A game, though, needs rules you can trust and a scene you can reproduce. Google warns that visual details and objects can drift during longer sessions, navigation or complex edits. This is a sketch that responds to the player, not a finished map to ship.
Official Google demonstration of Genie 4.
