- Google has released its new-generation world simulation model, “Genie 4,” through “Project Genie,” an experimental prototype available for “Google AI Ultra” members.
- “Genie 4” is built by integrating the capabilities of “Gemini Omni” (which replicates real-world visual behaviors and appearances) and various “Gemini” models (which handle complex reasoning to maintain overall consistency and continuity in simulations).
- Waitlist registration is now open on the “Project Genie” page for “Google AI Pro” members as well.
On Wednesday, November 18, 2026 (local time), Google released the new-generation world simulation model “Genie 4,” which was officially announced on Tuesday, October 27, 2026, via “Project Genie,” an experimental prototype for “Google AI Ultra” members.
“Genie 4” is an AI model built by integrating the capabilities of “Gemini Omni,” which reproduces real-world visual behaviors and appearances, and a group of “Gemini” models responsible for complex reasoning to maintain overall simulation consistency and continuity. Adopting a native multimodal design, it can simultaneously generate not only visual world simulations in real-time as scenes unfold, but also environmental sounds, background music, and character dialogue audio.
By using “Genie 4” in “Project Genie,” users can enter natural language prompts while exploring virtual spaces in their browsers to perform real-time environmental edits, such as “changing the weather,” “introducing characters,” or “altering the direction of actions.”

Alongside the rollout of “Genie 4,” this update also introduces the “Project Genie agent,” which allows users to build worlds by conversing with AI to automatically generate background music tailored to initial visuals and atmospheres. By integrating “Gemini” into an active exploration session, users can also receive assistance, such as observing screen conditions to suggest next actions or automatically triggering in-world events.
Furthermore, a new community gallery has been established where users can directly explore, remix, and publish official or community-created worlds from their browsers. Featured creations include “Living Memory” (which converts personal photos into explorable spaces), “Dino Field Guide” (a prehistoric habitat tour based on 360-degree panoramas), and collaborative works with creators such as “Villa Giallo” and “The Life of Ants.”
Additionally, the legacy model “Genie 3” remains available in the “Project Genie” model picker. To support longer explorations and spatial generation starting from 360-degree panoramic images, “Genie 3” has also been enhanced with this update as follows:
| Session Extension & World Linking |
|
|---|---|
| 360-Degree Panorama Grounding |
|
| Real-Time In-World Events |
|
All videos and audio generated in “Project Genie” are embedded with “SynthID” watermarking technology, and safety screenings are performed on all model inputs and outputs. The latest “Project Genie,” including “Genie 4,” is now rolling out to eligible “Google AI Ultra” members, and waitlist registrations are also open for “Google AI Pro” members on the “Project Genie” page.
Source: Google (https://blog.google/innovation-and-ai/models-and-research/google-deepmind/genie-4-now-available-project-genie/)





