What if building a video game world were as easy as describing it to a friend? Google is betting many people want to try.
On Wednesday, Google started rolling out Genie 4, its newest “world model,” to eligible Google AI Ultra subscribers through an updated Project Genie. A world model is an AI system that generates a 3D environment you can walk or fly around in real time. Genie 4’s big new trick is that you can change the world while you’re inside it. Type a prompt mid-session, and the weather shifts, a character appears, or the action heads somewhere else. It’s also the first Genie model with native audio, so you get ambient sound, music, and characters who talk back, all generated on the fly.
The older Genie 3 isn’t going anywhere, and it got upgrades too. Sessions now run up to three minutes instead of 60 seconds. You can chain worlds together into longer journeys, and you can start a world from a 360-degree panorama photo.
Google also added a Project Genie agent that builds a world from a plain conversation, creating the starting visuals and even a custom background score. There’s also a community gallery where people can publish their worlds and remix others' by changing the structure, soundtrack, or rules. Early showcase pieces include a room-by-room gothic manor mystery and a trip through an ant colony from an ant’s point of view.
There are limits. Access is only for Ultra subscribers for now, and AI Pro members are on a waitlist. Google admits it’s still working on lag, complex physics, precise audio, and keeping scenes consistent over time. Everything generated carries Google’s SynthID watermark.
