genie 4
Type a scene and Genie 4 builds a photoreal, explorable world with believable physics, ambient sound and live dialogue — all in real time.
Feedback
AI Dance Tools
Explore our AI-powered dance video creation tools
Muse Video
Seedance2.5
The Future of AI Video Is Here.
Seedance 2.0
The Future of AI Video Is Here.
Veo3.1
Create Stunning Videos with Veo3.1
Grok Video Generator
Create Videos from Text or Images with AI
MiniMax H3 video generator
MiniMax H3 AI Video Generator
Seedance 2.0
The Future of AI Video Is Here.
Gemini Omni 1.2 Flash
How Genie 4 Builds a Living World
Hand Genie 4 a sentence or a single still frame and it returns a world you can roam through — picture, physics, ambience and voices woven together in one pass.
- One Model for Sight, Sound and SpeechAmbience and spoken lines are produced together with the imagery, rather than being layered on once the Genie 4 world already exists.
- Steer the Scene in Plain LanguageCall for new weather, a fresh camera angle or another character mid-run, and Gemini reshapes the scene to match whatever is on screen.
- Powered by the Gemini Model FamilyGemini Omni keeps the visuals and motion anchored in reality, while companion Gemini models track continuity from one frame to the next.
How to Use Genie 4
Four steps take you from the waitlist to a scene you can walk around and steer.
What Genie 4 Can Do
A new class of world simulation, running live and reacting to everything you do.
Weight, Momentum and Collisions
Objects carry convincing mass and inertia, so aerodynamics, impacts and other interactions play out the way you would expect.
Convincing Human Motion
People in the scene walk, react and speak believably, and you can nudge them around while the world keeps running.
Sound That Belongs to the Scene
Audio is generated alongside the image, so every location carries its own atmosphere, from wind in the trees to breaking surf.
Dialogue That Answers Back
Characters respond to your inputs and to events inside the world as they happen, with speech that shifts as the scene changes.
Planning Across Many Steps
Gemini 3.8 Flash runs inside the loop — planning moves, inspecting frames and re-planning whenever the world shifts.
Cinematic, Photo-Grade Finish
Grain, lighting and lens behaviour stay consistent, so long takes and free camera moves still read as real photography.
Genie 4: Frequently Asked Questions
The questions people ask most about the Genie 4 world model, answered in plain language.
So what is Genie 4, exactly?
It is a natively multimodal world model: it predicts how a scene will look, sound and react as it unfolds.
How can I get access to Genie 4?
Access is limited to a research preview for vetted testers — sign up for the Project Genie waitlist, the Gemini API or AI Studio.
Can Genie 4 produce audio and voices?
It can. Ambient sound and interactive dialogue are generated by Genie 4 itself, and both shift in response to what you do.
How do I steer a running simulation?
Keyboard controls move you through the space, while text prompts redirect the evolving scene as it runs.
Where do the Gemini models fit in?
Gemini Omni anchors the visual style and behaviour, and the other Gemini models hold the world together across time.
What can Genie 4 be used for?
Teams use it for robot and agent training grounds, navigation benchmarks for AI, and entirely new kinds of interactive experience.
Get Early Access to Genie 4
Ask for access to Genie 4 through Project Genie, or try the early-access route via the Gemini API and Google AI Studio.