Google opens Genie 4 to developers through the Gemini API
Google is offering Genie 4 to developers through the Gemini API and AI Studio. The system generates interactive worlds in 720p at 24 frames per second with synchronized audio, which can be modified in real time using controls and text instructions.

Google now lets you try Genie 4 through the Gemini API and Google AI Studio, with a model that can generate interactive worlds from text, a reference image, or both.
The system creates continuous environments in 720p at 24 frames per second, with synchronized native audio. That includes ambient sounds, effects, and character dialogue. It does not just generate an image or a short video: it maintains a world that can change as you interact with it.
Worlds you can modify in real time
Developers can move through these environments using keyboard controls while sending text instructions at the same time. For example, they can change the following on the fly:
- The weather and lighting.
- Objects in the environment.
- Events taking place in the scene.
- How characters interact.
Gemini's reasoning helps maintain the consistency of the world's state. If you switch from a sunny day to a storm, the environment should reflect that change without forgetting what has already happened in the scene.
"Starting from a text prompt, a reference image, or both, developers can experiment with generating continuous 720p worlds at 24fps with synchronized native audio."
What you can use it for
The tool is designed for prototyping interactive experiences, visualizing spatial designs, and training robots or artificial intelligence agents in simulated environments.
A studio could describe a room and modify its layout before building it. A game development team could test a scene with different events. These worlds could also serve as practice spaces for an agent learning to navigate and act.
Google offers an interactive canvas and starter apps in AI Studio. To integrate it into their own products, developers must use the Gemini API through the Interactions API and consult the service's documentation and pricing.
Each result also includes an invisible SynthID watermark, Google's system for identifying AI-generated content without changing its appearance.
The important point is that Genie 4 does not present video generation as a finished result, but as an environment you can interact with and experiment in. The next thing to watch will be how much it costs to keep these scenes active and how well they maintain their coherence when instructions become more complex.

