Google launches Agentic video understanding for three Gemini models.
Google has launched Agentic Video Understanding for Gemini 3.7 Flash, 3.6 Flash, and 3.5 Flash-Lite, which allows users to dynamically browse video timelines and adjust frame rates and resolutions by question.
Google introduces Agentic Video Understanding for Gemini 3.7 Flash, 3.6 Flash, and 3.5 Flash-Lite.Unlike static processing that extracts frames per second at a fixed rate, this mode dynamically browses the timeline based on the issue, selectively checks subtitles, and adjusts the frame rate and resolution as needed.
This capability is now available through the Gemini API in Google AI Studio and the Gemini Enterprise Agent Platform, supporting file uploads and YouTube links. Google claims that in its tests, Gemini 3.7 Flash reduced token consumption by up to 88%, lowered costs by 66%, and improved accuracy by 7%; these figures are official test results, and actual performance depends on the materials and tasks.