genie 4

Turn one sentence or a still image into a playable world — Genie 4 simulates motion, sound and conversation together while you explore.

Genie 4 World Builder
Describe a scene, then step inside it
AI Video Prompt Generator

Feedback

AI Dance Tools

Explore our AI-powered dance video creation tools

Genie 4 in Depth: A World Model You Can Explore

Genie 4 takes a sentence or a reference frame and builds a scene you can move through — image, motion, sound and voices produced in one pass.

  • One Model for Sight, Sound and Speech
    Audio, ambience and spoken lines are produced in the same pass as the imagery, rather than layered on after the scene is finished.
  • Steer the Scene Mid-Run
    Ask for different weather, a swung camera or an extra character while the world is running, and Gemini reworks what is already on screen.
  • Powered by Gemini Reasoning
    Gemini Omni keeps appearance and behavior anchored in reality, while companion Gemini models track continuity from one frame to the next.

How to Start Simulating with Genie 4

Access, prompt, explore — the short path from the waitlist to a Genie 4 scene you can steer in real time.

Capabilities That Define Genie 4

Six things Genie 4 handles while the scene is still running in front of you.

Weight, Momentum and Impact

Objects carry believable mass, so collisions, airflow and falling debris move the way a viewer intuitively expects.

Convincing Character Motion

Figures walk, gesture and speak with natural timing, and you can redirect them on the fly as the world unfolds.

Sound Born With the Frame

Each setting arrives with its own soundscape, from wind in leaves to surf breaking on rock, produced alongside the picture.

Conversation That Answers Back

Characters reply to your prompts and to events inside the world, with speech that adapts as the scene develops.

Planning Across Many Steps

Gemini 3.8 Flash runs inside the loop, mapping out actions, reviewing frames and revising the plan when conditions shift.

Camera-Grade Visual Consistency

Lighting, grain and lens behavior hold steady across long takes and free camera moves, so every shot reads as photographed.

FAQ

Common Questions About Genie 4

Access, controls, audio and use cases for Genie 4 — the essentials answered in brief.

1

What is Genie 4 in simple terms?

It is a multimodal simulation model: hand it a prompt or an image and it predicts how that scene should look, sound and behave next.

2

How can I try it?

Access is gated during the research preview — request a spot on the Project Genie waitlist, or apply through the Gemini API and AI Studio.

3

Does it produce audio on its own?

Yes. Ambient sound and spoken dialogue come straight from the model and shift in response to what you do in the scene.

4

How do I steer a running scene?

Move with standard keyboard input, then type new instructions at any moment to redirect the camera, the weather or the cast.

5

Where do the Gemini models fit in?

Gemini Omni anchors realism in appearance and behavior, while other Gemini models maintain coherence as time advances.

6

What can I use it for?

Training grounds for robots and agents, navigation research, and interactive experiences that need a responsive environment.

Request Early Access to Genie 4

Add your name to the Project Genie list, or apply for Genie 4 preview access through the Gemini API and Google AI Studio.