Gemini Agentic Video Cuts Token Use by 88 Percent
Google published a post on its own blog.google titled "Introducing agentic video understanding with Gemini", by Rohan Doshi and Mario Lucic of Google DeepMind, dated September 1, 2026. Everything below comes from it.
Transcript
Google announced agentic video understanding on September first. Gemini now searches inside video instead of watching every frame.
Google says agentic video understanding launches across its latest Gemini models, searching frames, audio, and transcripts.
Google says static processing reads video at a fixed one frame per second. Gemini can now find one moment inside a long recording.
Google says costs drop up to sixty six percent, tokens up to eighty eight percent, and accuracy improves up to seven percent.
Google says it is available today in Google AI Studio with no extra feature fee. YouTube's Ask YouTube feature will use it in the coming months.
Inboxsmith builds AI phone and email agents for small business. Please like and subscribe for more news.
Sources
Every claim in this video comes from the top ranking coverage of this topic. The claims and where each one came from:
- Today, we’re launching agentic video understanding across our latest models: Gemini 3.7 Flash, 3.6 Flash and 3.5 Flash-Lite.(Google's official announcement, Introducing agentic video understanding with Gemini, published September 1, 2026)
- Unlike current ‘static’ processing, where the model ingests the video at a fixed frames-per-second rate (default 1 FPS, adjustable via API), agentic video understanding pairs the model’s core reasoning with native video tools to dynamically search, scan, and inspect target video segments across visual frames, audio, and transcripts.(Google's official announcement, Introducing agentic video understanding with Gemini, published September 1, 2026)
- Across standard video analysis benchmarks, Gemini models with agentic video understanding reduce analysis costs by up to 66% and token consumption by up to 88%, while improving accuracy by up to 7%.(Google's official announcement, Introducing agentic video understanding with Gemini, published September 1, 2026)
- These efficiency gains are especially pronounced on long-form video (from 10-minute how-to guides to 90-minute lectures and multi-hour recordings), where static processing forces developers to choose between high token costs or techniques that drop critical details.(Google's official announcement, Introducing agentic video understanding with Gemini, published September 1, 2026)
- Long-form needle-in-a-haystack search : Answer complex queries across multi-hour videos without consuming millions of tokens.(Google's official announcement, Introducing agentic video understanding with Gemini, published September 1, 2026)
- The feature is available today for video uploads and YouTube videos via the Gemini API in Google AI Studio and the Gemini Enterprise Agent Platform.(Google's official announcement, Introducing agentic video understanding with Gemini, published September 1, 2026)
- It uses standard Gemini API token pricing with no additional feature fee.(Google's official announcement, Introducing agentic video understanding with Gemini, published September 1, 2026)
- And in the coming months, agentic video understanding will also power YouTube's ‘ Ask YouTube’ feature on the video watch page, leveraging Gemini to deliver higher-quality answers grounded in the visuals.(Google's official announcement, Introducing agentic video understanding with Gemini, published September 1, 2026)