Google DeepMind's Genie 3 is a general-purpose "world model" — instead of generating a finished video clip, it builds an interactive 3D environment you can move through in real time, generated live from a text prompt or a single image. It's a different paradigm from linear video generators like Sora: less a camera, more a playable world.
This is a set of environments I made with Genie 3 during its early access release, testing what the model could do — a torch-lit crypt, an alpine ski run, a jungle dig, a shipwreck dive. The tool is early and experimental, but the quality of what comes out still depends on how you direct it: the same prompt-craft and visual judgment I've built across AI image and video work carries directly into steering a world model.