Direct answer: Google Genie 4 is not generally available yet. It is a research preview for trusted testers. Google says it is coming next to Google AI Ultra subscribers through Project Genie and to early-access partners through the Gemini API and Google AI Studio, but it has not published a general release date, public API model ID or API price.[2][4] The useful action today is to join the official Genie 4 waitlist, then prepare a small test brief while you wait.
Official Genie 4 links
| What you need | Official link | Current use |
|---|---|---|
| Request access | Genie 4 waitlist | Apply for the Project Genie rollout |
| Check rollout status | Google DeepMind’s Genie 4 announcement | Confirm whether access has expanded |
| Open the existing product | Project Genie | Check what your signed-in account can use |
| Read current product help | Project Genie help | Review account and usage requirements |
Do not pay for Google AI Ultra only to get immediate Genie 4 access. Google describes Genie 4 access for Ultra subscribers as “coming soon,” not available to every subscriber today.[4]
How to join the Google Genie 4 waitlist
- Open Google DeepMind’s Genie 4 launch page.
- Follow its Join the waitlist link to the Google-hosted application form.
- Sign in with an email address you monitor and complete only the information requested by the form.
- Describe a concrete use case if asked—for example, pre-visualizing a film scene, testing an interactive training environment or prototyping a game world.
- Save the submission confirmation. An application is not an invitation and Google has not stated an approval timetable.[4]
- Check Project Genie after any invitation arrives; do not assume the model is enabled merely because the Project Genie page opens.
Genie 4, Project Genie and Gemini are different
The names are easy to mix up:
- Genie 4 is Google DeepMind’s newly announced world model.
- Project Genie is the browser experience for creating and exploring generated worlds.
- Gemini is a broader family of Google AI models and products. Google says Genie 4 uses Gemini-family capabilities to help anchor and guide its simulations.[2][4]
Opening Gemini does not prove you have Genie 4. Likewise, access to the existing Project Genie experience does not prove your account has been upgraded to Genie 4.
Who can use Genie 4 right now?
As of October 28, 2026, Google describes three access stages:
- Trusted testers: the research preview is already rolling out to this limited group.
- Google AI Ultra subscribers: Genie 4 is coming to Project Genie, with a waitlist open now.
- Developers and partners: early access is planned through the Gemini API and Google AI Studio.[2][4]
Google has not published a broad-release date, a public API model identifier, general API pricing or rate limits for Genie 4. Avoid tutorials that provide an unverified model name or claim that a normal Gemini API key activates it.
A note about regions
Google’s public pages are not fully synchronized. A Google announcement says the existing Project Genie experience expanded gradually to eligible Google AI Ultra subscribers globally, while the current help page retrieved for this guide still says Project Genie is US-only.[5][6] Treat the signed-in product and the benefits listed for your own account and country as authoritative. This discrepancy also means you should not buy or change a subscription based only on an old tutorial.
What Genie 4 adds
Google’s demonstrations focus on four practical changes:
- Real-time natural-language steering: change weather, camera position, characters or objects while the simulation is running.
- Native generated audio: worlds can generate ambient sound, music and interactive dialogue alongside visuals.
- More realistic movement and physics: Google demonstrates aerodynamics, collisions and human actions.
- Gemini-guided events: Gemini models can break a task into steps, monitor frames and adjust a plan while Genie renders the world.[4]
Potential uses include rapid scene pre-visualization, interactive concept design, agent or robot training environments, education and scenario prototyping. These are early research uses, not proof that Genie 4 is ready to replace a game engine, simulator or production video workflow.
What you can do in Project Genie today
The existing Project Genie experience can create an environment from text or images, let you define a character and perspective, and generate an explorable world in real time. Google has also announced Street View grounding for the existing product.[5][6]
Google’s current help material describes requirements for the existing experience rather than guaranteed Genie 4 access. It lists an adult user, a personal Google account and a Google AI Ultra subscription, but its regional wording conflicts with Google’s later global expansion announcement.[5][6]
If Project Genie is available in your account, use it to learn the workflow now:
- Start with one small environment rather than a whole game.
- Define the environment, character, camera perspective and one action.
- Generate the world and test movement.
- Note where objects, physics or visual details drift.
- Save the prompt and reference assets so you can rerun the same test when Genie 4 reaches your account.
Copy-paste Genie world brief
Use this planning template while waiting for access:
PURPOSE:
Create a short interactive prototype for [film / game / training / education].
ENVIRONMENT:
[Location, time of day, weather, materials, lighting and visual style.]
CHARACTER OR VEHICLE:
[What the user controls, appearance and movement style.]
CAMERA:
[First person / third person / fixed / moving; lens and framing.]
USER ACTION:
[One primary action such as walk, drive, fly, inspect or move an object.]
WORLD RESPONSE:
[What should change when the user acts.]
AUDIO:
[Ambient sound, music, dialogue and events that trigger sound.]
LIVE EDIT TO TEST:
[Change weather / add an object / move the camera / introduce a character.]
CONSISTENCY CHECK:
[Objects, signs, character appearance or physics that must remain stable.]
SUCCESS CRITERIA:
[What a useful two-minute test must demonstrate.]
Keep the first test narrow. A single measurable interaction makes it easier to distinguish genuine model improvement from a visually impressive demo.
Current limitations and safety markers
Google says Genie 4 remains experimental. Visual details and object permanence can drift in longer sessions or after multiple complex edits. Complex physical interactions, accurate real-time audio, legible text, spatial accuracy and long-term consistency remain active research areas.[4]
Google also says all video and audio produced by Genie 4 will include SynthID, its imperceptible watermark for identifying AI-generated media.[4] If you eventually use outputs publicly, label them as AI-generated and do not present a simulated event as documentary footage.
Should you join the waitlist?
Join it now if you work in game concepts, film pre-visualization, interactive education, simulation, robotics or agent training and can describe a specific test.
Wait for public documentation if you need a production API, contractual availability, stable quotas, predictable pricing or a published model ID. None of those have been announced for general developers yet.[4]
Frequently asked questions
Is Google Genie 4 available to everyone?
No. It is currently a research preview for trusted testers. Google says access is coming to Google AI Ultra subscribers in Project Genie and to early-access API and AI Studio partners.[4]
How do I access Google Genie 4?
Use the waitlist linked from Google DeepMind’s official Genie 4 page. Applying does not guarantee access or provide an activation date.
Do I need Google AI Ultra?
Google names Google AI Ultra subscribers as the upcoming consumer audience through Project Genie. However, buying Ultra does not mean Genie 4 is active immediately; the launch page still describes that rollout as coming soon.[4]
Is there a Genie 4 API?
Google says early-access partners will receive access through the Gemini API and Google AI Studio, but it has not published general API documentation, a public model ID or pricing.[2][4]
Is Project Genie the same as Genie 4?
No. Project Genie is the product experience; Genie 4 is the new model. Existing Project Genie access does not by itself confirm Genie 4 access.
Can Genie 4 generate audio?
Yes. Google says it generates native audio, including ambient sounds, music and interactive dialogue, alongside the simulated world.[2][4]
What are Genie 4’s main limitations?
Google reports possible visual drift, changing object permanence, imperfect complex physics and audio, difficulty rendering legible text, spatial inaccuracies and limited long-term consistency.[4]