Google began rolling out Genie 4 on November 18 through Project Genie to eligible Google AI Ultra subscribers, letting users change generated worlds while exploring them in a browser. The model also introduces native audio generated in real time, with sounds that adapt to events within a scene. Access comes with a broader Project Genie update that includes improvements to Genie 3.
Users can type instructions during a session to alter their surroundings, introduce a character or change how events play out, Google says. Audio can range from environmental sounds to dialogue between characters, responding as the world changes. These controls follow feedback from early testers who wanted more influence over events and simpler ways to create and share worlds.
The rollout follows Google's October introduction of Genie 4 to a small group of trusted testers, when subscriber access through Project Genie was still promised for later. During Project Genie's earlier public testing, access required a $249.99 monthly Google AI Ultra subscription, and exploration was limited to 60 seconds per world. That earlier prototype used Genie 3 and lacked the demonstrated ability to add elements while a world was already running.
The updated experience also introduces the Project Genie agent, which Google says helps users build the worlds they envision. A community gallery lets people find worlds made by others, explore them and create remixed versions. Project Genie combines Google's world models with Gemini models to let users generate, enter and modify those environments through their browser.













