Google DeepMind announced that Genie 4 is now part of Project Genie for Google AI Ultra subscribers. The new model lets users change a virtual world while they explore it. Just type a new prompt and the scene updates instantly – you can switch the weather, add characters or rewrite the story on the fly. Genie 4 also brings native audio that reacts to the changes, from background sounds to spoken dialogue. The update also includes upgrades to Genie 3, a community gallery and a Project Genie agent that helps you build worlds through conversation. Genie 3 now supports three‑minute sessions (up from one minute) and can chain multiple worlds together. It can also seed new worlds from 360‑degree panorama images. The agent coordinates Google’s multimodal models, including Gemini, to suggest next steps or trigger events as you move through a world. A gallery of curated experiences is now live. It features personal‑photo spaces, a dinosaur field guide built from real panoramas, and a rock‑paper‑scissors arena where Gemini acts as a referee. Early collaborators like creative technologist Will aka BLVCKL!GHT and filmmaker Anand Gandhi have demonstrated what Genie 4 can do, showing a gothic‑manor mystery and an ant‑colony adventure. Google says the rollout is limited to current Google AI Ultra subscribers, with Google AI Pro members able to join a waitlist. All generated video and audio carry a SynthID watermark for safety, and inputs and outputs pass through safety screenings.
How Genie 4 fits into Google’s AI roadmap
Project Genie combines Google’s world‑model research with its Gemini large‑language model (LLM) – the type of AI behind ChatGPT. The goal is to let users create, explore and remix immersive environments directly in a browser. Genie 4 is described as the most realistic and interactive world model to date, but Google notes ongoing challenges with camera latency, complex physics and long‑term scene stability. A research overview of Genie 4 is linked for those who want technical details.
Why it matters
For people with a Google AI Ultra subscription, Genie 4 means they can design and modify immersive scenes without coding, using simple text prompts. It also opens a shared space where anyone can explore or remix community‑created worlds, turning personal photos or ideas into interactive experiences.