Google DeepMind has begun rolling out Genie 4, its most advanced world model, to eligible Google AI Ultra subscribers through an updated Project Genie experience. The model lets users type new prompts while they explore a generated world, changing the weather, adding characters or redirecting the action in real time. Genie 4 also pairs these visuals with native audio that adapts to the scene, from ambient soundscapes to character dialogue. Google AI Pro members can join a waitlist.
The update also upgrades the earlier Genie 3 model, which remains available in the model picker. Genie 3 now supports runs of up to three minutes per generation, up from 60 seconds, and lets users chain multiple worlds into longer journeys. It can also seed new worlds from 360-degree panorama images. A new Project Genie agent builds worlds from natural-language conversation, using Google's multimodal models to create the initial visuals and background music. During a session, Gemini can observe the screen, suggest where to go next, or trigger in-world events. Creators can publish their worlds to a new community gallery for others to explore and remix. The gallery also includes curated examples such as 'Living Memory', which turns personal photos into explorable spaces, and 'Villa Giallo', a room-by-room mystery built with Genie 4.
For Australian businesses and professionals, this changes little today. Access is limited to the paid Ultra tier, and the product is a consumer and creative tool rather than enterprise software. It is worth a look for marketing, design or training teams who want to prototype interactive scenes, but staff should not upload client photos, customer data or other personal information to a public gallery without checking the Privacy Act 1988 obligations and the platform's terms first.
Key Points
Genie 4 is rolling out to eligible Google AI Ultra subscribers, with Pro members able to join a waitlist.
Users can change environments mid-session with text prompts, and Genie 4 adds native audio that adapts to the scene.
Genie 3 now runs for up to three minutes per generation and supports linked worlds and 360-degree panorama seeding.
A new Project Genie agent builds worlds through conversation and can bring Gemini into a session to suggest or trigger events.
A community gallery lets creators publish, share and remix worlds.