Google launches agentic video understanding for Gemini models

Google has introduced agentic video understanding for its Gemini 3.7 Flash, 3.6 Flash, and 3.5 Flash-Lite models, enhancing video analysis efficiency and accuracy. The feature allows models to dynamically scan video segments, reducing token consumption by up to 88% and costs by up to 66%, while boosting accuracy by up to 7%. Available today via the Gemini API in Google AI Studio and the Gemini Enterprise Agent Platform, agentic video understanding leverages native video tools to improve performance and unlock capabilities like sub-second moment retrieval and precise counting. Unlike static processing, which uses a fixed frame rate, agentic video understanding enables models to actively determine what to watch, at what speed, and through which modality, significantly reducing development overhead. Early access partners reported strong performance, with Gemini 3.7 Flash positioned at the accuracy-to-cost pareto frontier for video analysis. The feature will soon roll out to all Gemini app users and power YouTube’s ‘Ask YouTube’ feature, delivering higher-quality answers grounded in visuals.