From the archive
Gemini 2.0 Flash launches as an experimental model
Google opens Gemini 2.0 Flash for multimodal input and text output, with a new live API and limited output previews.
Google DeepMind introduced Gemini 2.0 Flash on December 11, 2024. The experimental model reached the Gemini API through AI Studio and Google Cloud Vertex AI , as well as a selectable mode in the Gemini app on desktop and mobile web.
Multimodal input and tools
The public developer release accepted text, images, video, and audio and returned text. The model also supported calls to tools such as Google Search, code execution, and application-defined functions.
Google demonstrated native image and speech output, but those output capabilities initially went to early-access partners. The broader release was still experimental.
A live API for interactive applications
The announcement also introduced the Multimodal Live API for real-time audio and video-stream input with tool use. Separate research demonstrations included Project Astra, Project Mariner, and Jules. Those prototypes showed work in progress rather than a general release of every demonstrated agent feature.