Google DeepMind rolls out Genie 4 world model with real-time audio and prompt edits

Google AI Ultra subscribers can now change weather, add characters or redirect action mid-session by typing; Genie 4 adds adaptive native audio for the first time.
Why it matters
Genie 4 moves Google's world models from a research preview toward a paid consumer product: worlds can now be edited while you move through them, and a new Project Genie agent uses Gemini and other Google models to build them through conversation. Users can publish and remix worlds in a community gallery. Google still calls Genie 3 and Genie 4 frontier research, citing camera latency, complex physical interactions, audio precision and long-term scene permanence as open problems. Watch whether access widens beyond the Ultra tier and how the waitlist for Google AI Pro members moves.
Key facts
- Available from 18 Nov 2026 to eligible Google AI Ultra subscribers
- Google AI Pro members can join a waitlist
- Worlds can be edited mid-session by typing a new prompt
- First Genie model with real-time native audio
- Genie 3 stays in the model picker for longer runs
- All generated video and audio carries SynthID watermarks