Google DeepMind · October 27, 2026 · Research preview
Gemini Genie 4: DeepMind's Genie 4 world model
DeepMind's October 27, 2026 post announces Genie 4, a world model for real-time interactive simulation with native audio. People also type Gemini Genie 4 because the post says Gemini models help keep the simulation coherent. Nano AI does not run Genie 4.
This generator runs Veo 3.1 Fast for 60 credits a clip. It makes a short video from a prompt. It is not Gemini Genie 4, and it does not steer a world in real time.
Ratio
Generated Videos
No videos yet
Fill out the form and click Generate Video to create your first AI video.
Tips: Use detailed descriptions for better results
- Made by
- Google DeepMind
- Released
- October 27, 2026
- Run on Nano AI
- Veo 3.1 Fast · from 60 credits per video
- Page updated
- October 31, 2026
Source: deepmind.google
What Gemini Genie 4 refers to
Google DeepMind's blog on October 27, 2026 names the model Genie 4, not a separate product called Gemini Genie 4. The same post says Genie 4 builds on the Gemini family: Gemini Omni anchors how a simulation looks and behaves, and other Gemini models help keep it coherent. A later section says Gemini 3.8 Flash can sit in the loop, watch frames, and break a goal into steps. Those are DeepMind's descriptions of Genie 4. They are not controls on this website.
The model page on deepmind.google says you can change a running simulation with language: weather, camera, a new character or object. It also says Genie 4 is their first real-time world model that generates audio itself, including ambient sound and spoken lines that react while the simulation runs. The blog says a research preview has started for trusted testers, with a later rollout planned for Google AI Ultra subscribers in Project Genie and for early-access partners of the Gemini API and Google AI Studio. The waitlist, if you want it, is the one DeepMind links from that blog. This site has no waitlist and no Genie API.
DeepMind also lists limits: details and object permanence can drift as a session runs longer, after several complex edits, or while you move through the scene. Complex physics, accurate audio, readable text, spatial accuracy and long consistency are still research. The model page says interaction lasts a few minutes, not an extended session, and that real places are not reproduced perfectly. Text is often legible only when the wording was in the description of the world.
Gemini Genie 4 at a glance
- Official name
- Genie 4, from Google DeepMind
- Announced
- October 27, 2026, on the DeepMind blog
- What it is
- A real-time world model, not a clip button
- Audio
- Native audio, including ambient sound and dialogue, in DeepMind's description
- Gemini's role
- Gemini Omni plus other Gemini models, including Gemini 3.8 Flash as a guiding agent in their write-up
- Who can use it
- Research preview for trusted testers; a wider rollout is described as still ahead
- On Nano AI
- Veo 3.1 Fast, 60 credits a clip. Not Genie 4
- What the clip is
- A short generated video. No live steering, no Project Genie session
How to read a Gemini Genie 4 result
The name mixes two products
Genie 4 is the world model. Gemini is the model family DeepMind says it uses inside that model. A Gemini image or video model on another site is not Genie 4.
A clip is not a world you can steer
Veo 3.1 Fast returns one video and stops. DeepMind's Genie 4 description is a session you can change while it runs. This page cannot start that session.
Treat the limits as part of the product
DeepMind says objects can drift, text is unreliable, and a session is a few minutes. A demo reel is not a promise that a long scene stays consistent.
Use your own scene
The prompts below describe places and objects. They do not ask the model to copy a film, a game, or a real person's likeness.
Genie 4 and the clip model on this page
The Genie 4 column quotes Google DeepMind's blog and model page as read on October 31, 2026. The Veo column is what Nano AI actually runs.
| Genie 4 | Veo 3.1 Fast on Nano AI | |
|---|---|---|
| Kind | Real-time world model | Text-to-video clip |
| While it runs | Language can change weather, camera and objects | The prompt is fixed, then one clip comes back |
| Audio | Generated with the world, per DeepMind | The clip can include sound; it is not a Genie session |
| Access | Trusted-tester preview | Available in the generator above |
| Price here | Not sold on Nano AI | 60 credits a clip |
How to make a short clip on this page
- 1
Stay on Veo 3.1 Fast
The generator above is already set to Veo 3.1 Fast. Leave it there if you want the 60-credit clip. Switching models changes the price.
- 2
Pick a scene
Use one of the scene buttons or edit the prompt. Describe the place, the light and the camera move. Do not write a film title or a real person's name.
- 3
Generate one clip
Sign in. The credit cost is shown before the clip starts. Download the file if you want to keep it. It will not continue as an interactive world.
Scenes for a short clip
These prompts are for Veo 3.1 Fast on this page. They do not start a Genie 4 session.
Rainy harbor
A quiet harbor at dusk after rain, wet stones, one empty wooden boat, gulls lifting off, cool blue light, the camera glides forward slowly, ambient water and wind, no people, no readable text.
Paper theatre
A handmade paper theatre of a mountain road, layered card hills, a paper bus with no logos, warm desk-lamp light, the camera pushes gently through the set, soft paper rustle, no people, no readable text.
Morning kettle
A kitchen table in morning light, a steel kettle begins to steam, a ceramic cup, a linen cloth, the camera holds still, soft room tone and a kettle whistle, no people, no readable text.
Clay fox
A small clay fox walks across a mossy log in a quiet forest, soft green light, shallow depth of field, smooth motion, leaves shift in a light wind, no people, no readable text.
FAQ
Gemini Genie 4 FAQ
Short answers from DeepMind's October 27, 2026 posts, and what this website actually runs.
Is Gemini Genie 4 a Gemini model I can call?
DeepMind's announcement calls the world model Genie 4. It says Gemini models help the simulation. It does not publish a Gemini API model ID named Gemini Genie 4 on the pages this article quotes. Nano AI does not offer a Genie 4 endpoint.
Can I steer a world on Nano AI?
No. The generator on this page runs Veo 3.1 Fast and returns one clip for 60 credits. It does not keep a live session, change the weather mid-run, or open Project Genie.
Does Genie 4 have sound?
DeepMind says Genie 4 generates audio with the world, including ambient sound and dialogue that reacts during the simulation. That claim is about Genie 4, not about the Veo clip on this page.
Who can use Genie 4 today?
DeepMind says a research preview is open to trusted testers, and that a wider release to Google AI Ultra subscribers in Project Genie and to early-access Gemini API and AI Studio partners is still ahead. Check DeepMind's blog for the current list.
What are the published limits?
DeepMind says visual details can drift, real places are not exact, on-screen text is often unclear unless it was in the prompt, and continuous interaction is a few minutes rather than a long session.
Is Nano AI affiliated with Google or DeepMind?
No. Nano AI is an independent service. It is not affiliated with Google, Google DeepMind, Gemini or Project Genie.
More from Nano AI
Related pages
Nano AI is an independent service and is not affiliated with Google, Google DeepMind, Gemini, Genie or Project Genie. Gemini Genie 4 is a search phrase for Genie 4. Facts about Genie 4 are from DeepMind's blog and model page as read on October 31, 2026. The generator on this page runs Veo 3.1 Fast.