Google Gemini's new agent-based video analysis cuts token usage by up to 88 percent
88% token cut. Google's Gemini now decides which video frames matter—not you.
Why it matters
Google is shipping agent-based video analysis as a product feature across three Gemini Flash models, meaningfully reducing inference costs and improving accuracy on long-form video. Practitioners using Gemini for video workloads can expect immediate ROI on token spend; this is a rare case where autonomy (frame selection) becomes a cost lever.
The key facts
5 to knowAgent-based video analysis deployed across Gemini 3.7 Flash, 3.6 Flash, and 3.5 Flash-Lite
Up to 88% token usage reduction reported
Model autonomously selects which segments and resolutions to examine (vs. fixed frame-by-frame scanning)
Accuracy improvements noted, especially on multi-hour footage
Feature is live as a product update
Go to the source
The Decoderthe-decoder.com
Publisher excerpt: Google is adding agent-based video analysis to Gemini 3.7 Flash, 3.6 Flash, and 3.5 Flash-Lite. Instead of scanning videos frame by frame at a fixed rate, the model decides on its own which segments to examine and at what resolution. Google says this cuts token usage by up to 88 percent while…