Google Deepmind is releasing Genie 4, its most advanced world model for generating interactive 3D environments, to Google AI Ultra subscribers. This update moves the technology from a research preview to a paid tier where users reshape worlds through text input in real time. The model generates native real-time sound that adapts to the scene. Older Genie 3 instances receive updates including longer sessions of up to three minutes and 360-degree panorama support. A new AI agent helps build worlds from spoken descriptions while a community gallery lets users share and remix creations. All generated content carries SynthID watermarks and goes through safety checks. Google AI Pro members can join a waitlist for access.
The shift to a paid subscription model means these tools are now restricted to specific tiers rather than being publicly available. The introduction of chainable worlds allows users to extend narrative sequences beyond previous limits. Native audio generation reduces the need for external sound libraries when building scenes. The community gallery facilitates sharing of user-created content within the platform. Safety checks and watermarks ensure all outputs are tracked and screened before distribution.
* Access requires a Google AI Ultra subscription.
* Sessions can now last up to three minutes.
* Content is watermarked with SynthID.


