On November 18 Google DeepMind said Genie 4, the world model it previewed to trusted testers on October 27, is now available to eligible Google AI Ultra subscribers in a new Project Genie experience. Users can type prompts mid-session to change the weather, add characters or redirect the action, with native audio generated alongside the visuals. A new Project Genie agent builds worlds through conversation, and a community gallery lets people publish and remix worlds. Genie 3 also gets longer sessions, chained worlds and 360-degree panorama grounding. Google AI Pro users can join a waitlist; no quantitative benchmarks were published.
Key Takeaways
- ✓Availability: eligible Google AI Ultra subscribers (18+) can use it starting today; Google AI Pro users can join a waitlist, and daily usage limits apply
- ✓Genie 4: type a new prompt mid-session to edit the world in real time, with native audio such as ambient sound and character dialogue generated alongside the visuals
- ✓Project Genie agent: describe your idea in plain language and the agent orchestrates Google multimodal models for opening visuals and a score; Gemini can watch the screen, suggest next steps and trigger events
- ✓Community gallery: browse, publish and remix worlds, with debut collaborations such as Villa Giallo with BLVCKL!GHT and The Life of Ants with filmmaker Anand Gandhi
- ✓Genie 3 upgrades: 3-minute sessions, chained worlds and 360-degree panorama grounding; all outputs carry SynthID, and no Gemini API date was given

Key Decision Metrics at a Glance
Turn your technical choice into a development budget
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Background
World models generate environments you can explore and interact with in real time, but until now Genie 4 was limited to a small group of trusted testers. On November 18 Google DeepMind announced on X and its official blog that Genie 4 is now in Project Genie, its first release to paying users, following through on the October 27 preview's promise that it was coming to Google AI Ultra.
How it works
Per the blog and model card, Genie 4 builds on the Gemini family, with Gemini Omni anchoring real-world appearance and behavior. The new experience has three parts. First, you can type prompts mid-session to edit the world in real time, with ambient sound, music and character dialogue generated alongside it. Second, the Project Genie agent turns a conversation into a world by orchestrating Google multimodal models for opening visuals and a score, and Gemini can watch the screen, suggest next steps or trigger events. Third, a community gallery supports publishing and remixing. Genie 3 also gets 3-minute sessions, chained worlds and 360-degree panorama grounding.
Numbers that matter
Google has not published quantitative benchmarks, resolution or frame rate. It says camera latency, complex physical interactions, real-time audio accuracy and long-term scene consistency are still open problems.
What it means for developers
Eligible Ultra subscribers aged 18+ can sign in at labs.google/fx/projectgenie, with daily usage limits; Pro users can join the waitlist. All outputs carry a SynthID watermark. Google has not announced timing or pricing for the Gemini API and AI Studio, so for now developers can only try it in the web app.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.