November 18, 2026 — Google is rolling out a new Project Genie experience to eligible Google AI Ultra subscribers, giving them first-time access to Genie 4, its most realistic interactive world model yet. Why it matters: a frontier world model that was limited to trusted testers in October is now in the hands of paying subscribers, inside a product built for creating and sharing explorable worlds.
What happened
Google announced the rollout on its official blog on November 18. The updated Project Genie combines first-time access to Genie 4 with upgrades to Genie 3, a new Project Genie agent that helps users build worlds conversationally, and a community gallery where people can discover, explore and remix worlds made by others. Eligible Google AI Ultra subscribers can try it starting today, and Google AI Pro members can join a waitlist on the same page.
What Genie 4 adds
Genie 4 lets users change a world while they are inside it. Instead of starting over, you can type a new prompt mid-session to shift the weather, add a character or redirect what is happening. Google says Genie 4 also pairs its visuals with real-time native audio that adapts as a scene unfolds, from ambient sound to character dialogue. Google positions Genie 4 as its highest-fidelity option for visuals, physics and native audio.
Genie 3 is not going away. It stays in the model picker for longer runs and for worlds seeded from 360-degree panorama images, and it gains longer sessions of up to 3 minutes per generation, up from 60 seconds, plus the ability to link multiple worlds together. The new Project Genie agent goes beyond a single prompt: it orchestrates Google’s multimodal models to set a world’s look and add a custom background score, and Gemini models inside a session can surface suggestion chips or trigger in-world events.
Why it matters
World models are moving from research demos toward products people can actually subscribe to. That matters for games, prototyping, training environments for robots and agents, and anyone testing ideas in a simulated space before building the real thing. Google is also being plain about the limits: Genie 3 and Genie 4 remain frontier research systems, and the company lists camera latency, intricate physical interactions, real-time audio precision and long-term scene permanence as open problems. Every video and audio output in Project Genie carries SynthID watermarking, and inputs and outputs go through safety screening. Our take, in the spirit of everything AI, tested: the access expansion is real and significant, but the permanence problem is the one to watch — a world that cannot reliably remember itself is a demo, not yet a workplace.
Frequently asked questions
Who can use Genie 4 now?
Eligible Google AI Ultra subscribers can use the new Project Genie experience starting November 18, 2026. Google AI Pro members can join a waitlist.
What is new in Genie 4 compared with Genie 3?
Genie 4 adds mid-session prompting to edit a world in real time and real-time native audio. Genie 3 remains available and now supports sessions of up to 3 minutes, world chaining and 360-degree panorama seeding.
Is Genie 4 a finished product?
No. Google describes Genie 3 and Genie 4 as frontier research world models and names camera latency, complex physics, audio precision and long-term scene permanence as unresolved challenges.
Sources: Google (official blog, Project Genie update featuring Genie 4, Nov 18, 2026), Unite.AI

