Agentic Video Understanding in Gemini
Agentic video analysis for faster, smarter Gemini insights
236
Product Hunt Upvotes
Sep 6, 2026
Launch Date
4d ago
First Tracked
About
AI Summary
Agentic Video Understanding in Gemini is a processing mode that allows the AI to autonomously select viewing parameters for videos, significantly reducing token usage by up to 88% and costs by up to 66%, while improving accuracy by up to 7%, particularly for long-form content. It is available through the Gemini API in AI Studio and the Gemini Enterprise Agent Platform at standard pricing.
Agentic video understanding is a new Gemini processing mode (3.7 Flash, 3.6 Flash, 3.5 Flash-Lite) that lets the model decide what to watch, at what speed, and through which modality, instead of a fixed frame rate. Cuts tokens by up to 88%, cost by up to 66%, boosts accuracy up to 7%, biggest wins on long-form video. Live now via Gemini API in AI Studio and Gemini Enterprise Agent Platform, just set processing to "agentic," standard pricing, no extra fee.
Tags