Google on Thursday made its Genie 4 model available to developers through the Gemini API and Google AI Studio. The company said the release follows the launch of Genie 4 a day earlier.
According to Google, developers can start from a text prompt, a reference image, or both. The model then generates continuous worlds at 720p resolution and 24 frames per second, which is the number of images shown each second. Google said the worlds come with synchronized native audio, including ambient sound, sound effects and character dialogue.
Google said developers can use real-time keyboard controls to move through the environments. They can also type live text prompts to change weather, lighting, objects or events as the world runs. The company said Gemini reasoning keeps the world state consistent and coherent throughout.
Google named three possible uses: prototyping interactive experiences, visualizing designs for spatial concepts, and training robotics and AI agents. The company described Genie 4 as a blank canvas for testing ideas.
Google said every output includes imperceptible SynthID watermarking. SynthID is Google’s technology for marking content it generates. The announcement did not explain how the watermark works.
Access comes through the Gemini API by way of what Google calls the Interactions API. Google AI Studio offers an interactive canvas and starter apps, according to the company. The announcement points developers to separate pages for pricing details and API documentation. It did not state the price in the post itself.
The announcement did not say how long each generated world can run, or whether usage limits apply.