Google DeepMind is rolling out Genie 4, its latest interactive world-generation model, to eligible Google AI Ultra subscribers as part of an updated Project Genie experience1.
The model lets users modify environments in real time through natural-language text prompts, including changing weather, adding characters, and triggering adaptive native audio. Diego Rivas, Group Product Manager at Google DeepMind, and Pavan Kumar, Group Product Manager at Google Labs, are listed as the authors of the announcement, published on November 18.
Alongside Genie 4, Google is shipping the Project Genie agent, a companion tool that helps users build custom worlds through natural-language conversations. The updated Genie 3 model also receives improvements: longer sessions, panorama grounding, and support for linked world sequences.
A new Project Genie gallery allows users to publish their creations for others to explore and remix. Google AI Pro members who do not yet have access can join a waitlist.
ANALYSIS Gating Genie 4 behind Google AI Ultra, Google's highest-tier consumer AI subscription, ties a generative-world capability directly to subscription revenue rather than offering it as a free research preview.