Genie 4: Google DeepMind's World Model, Waitlist & How to Try
Genie 4 launched Oct 27, 2026: real-time worlds with native audio, live text steering and 2-minute scenes. What it is, how to join the waitlist and who gets access.
Quick answers
- What is Genie 4?
- Google DeepMind's newest world model, announced Oct 27, 2026. It generates photorealistic, interactive 3D worlds in real time, with native audio and dialogue, that you can explore and change with text prompts.
- Can I try Genie 4?
- Not openly yet. It is a research preview for trusted testers. You can join the Genie 4 waitlist for Project Genie on the DeepMind blog.
- Who gets Genie 4 first?
- After trusted testers, Google AI Ultra subscribers in Project Genie, then early access partners in the Gemini API and Google AI Studio.
- What is new vs Genie 3?
- Native audio and speech, real-time text editing of the scene, better physics and people, and consistent worlds for about two minutes, double Genie 3.
- Is Genie 4 free?
- Google has not announced pricing. Consumer access is planned through the paid Google AI Ultra plan.
Google DeepMind announced Genie 4 on October 27, 2026, a little over a year after Genie 3. Genie 4 is a real-time world model: type a prompt or drop in an image, and it builds a world you can walk through, steer and listen to. It is DeepMind's most realistic Genie yet and the first to generate sound natively.
It is still an early, experimental model. Access starts with trusted testers; everyone else can join the Genie 4 waitlist.
1. Genie 4 at a glance
Item | Genie 4 |
|---|---|
Announced | Oct 27, 2026 |
Developer | Google DeepMind |
Type | Real-time interactive world model |
Inputs | Text prompts, reference images, keyboard controls |
Output | Explorable video world with audio and dialogue |
Scene consistency | About 2 minutes (Genie 3: about 1 minute) |
Built on | Gemini Omni; Gemini 3.8 Flash as guiding agent |
Status | Research preview, trusted testers |
Next access | Google AI Ultra (Project Genie), Gemini API, AI Studio |
2. What Genie 4 can do
DeepMind lists four key capabilities, all shown in real-time recordings where users move with standard keyboard controls and type prompts to steer the scene.
Steer the world in real time
Change the weather, move the camera, or add a person or object mid-simulation with plain language. Gemini rewrites your prompt to fit the scene and, between inputs, keeps the world moving forward coherently.
More realistic physics and people
Genie 4 simulates aerodynamics, collisions and other physics, and its biggest jump is people: characters move naturally, react to their surroundings and can be steered live. Demos include a crowded 1980s Japanese train station and a watermelon smashed in slow motion.
Native audio and dialogue
Every world carries its own ambient sound, from wind to waves, plus music and interactive dialogue: characters talk and react to you or to events in the world, in real time. Genie 3 was silent.
Complex events with Gemini
Gemini 3.8 Flash runs as an agent in the loop. It watches each frame, breaks goals into steps and checks that each step happened, which gives Genie 4 more reliable cause and effect for multi-step scenes, such as a robot arm pipetting liquid into a lab plate.
3. Genie 4 vs Genie 3
Genie 3 (2025) | Genie 4 (Oct 2026) | |
|---|---|---|
Audio | None | Native ambient sound, music, dialogue |
Consistent scene length | About 1 minute | About 2 minutes |
Editing | Promptable world events | Freeform live edits: weather, camera, new people or objects |
People and physics | Basic | More human-like characters, better physics |
Reasoning | Model only | Gemini 3.8 Flash guiding agent |
Public access | Project Genie for AI Ultra | Waitlist; AI Ultra and Gemini API next |
4. How to try Genie 4: waitlist and access
- Join the waitlist: use the Genie 4 waitlist form linked from the DeepMind announcement to request Project Genie access.
- Google AI Ultra: Ultra subscribers are first in line for Genie 4 inside Project Genie, the same app that hosted Genie 3.
- Developers: early access partners will get Genie 4 in the Gemini API and Google AI Studio; there is no public model ID yet.
- Trusted testers: DeepMind is already working with cohorts in creative industries, journalism, education, gaming and engineering.
- Today: watch the official demo video on the Google DeepMind YouTube channel.
Google has not announced a public release date or price. Expect the rollout to follow Genie 3's path, which moved from research preview to Ultra subscribers over several months.
5. What Genie 4 is for
DeepMind frames world models as a step toward AGI: safe, endlessly varied training grounds for robots and AI agents. The nearer-term uses are creative: previs for film, interactive journalism, education, game prototyping and manufacturing simulations. Rivals are moving fast too; Runway shipped GWM Worlds 2, and World Labs is being acquired by AMD.
6. Where Sumus fits
Yes, you can start building with Google's models today while you wait for Genie 4. Sumus puts Gemini, Claude, GPT and other top models into one team chat as AI employees, so you can draft Genie prompts, storyboard worlds and compare outputs from several models in one place.
What we're watching
- Genie 4 arriving for Google AI Ultra subscribers in Project Genie.
- A Genie 4 model in the Gemini API and AI Studio, and its pricing.
- Longer scene lengths beyond two minutes.
FAQ
When was Genie 4 released?
Google DeepMind announced Genie 4 on October 27, 2026, as a research preview for trusted testers.
How do I get access to Genie 4?
Join the waitlist linked from the DeepMind blog. Google AI Ultra subscribers and Gemini API early access partners are next.
Does Genie 4 have sound?
Yes. It is DeepMind's first real-time world model with native audio, including ambient sound, music and interactive dialogue.
How long can a Genie 4 world last?
DeepMind and press reports say worlds stay consistent for up to about two minutes, double Genie 3.
What model powers Genie 4?
Genie 4 builds on the Gemini family: Gemini Omni for realism and Gemini 3.8 Flash as a guiding agent.
Is there a Genie 4 API?
Not publicly. Early access partners will get it in the Gemini API and Google AI Studio first.
Sources
- Google DeepMind: Genie 4, a new generation of world simulation, Oct 27, 2026
- Google Keyword blog, The Verge and TechCrunch coverage, Oct 27, 2026. Read Oct 28, 2026.
Update log
- — Oct 28: First published. Genie 4 announced Oct 27, 2026; capabilities and access from Google DeepMind.