Gemini's Agentic Video Understanding Cuts Tokens 88%, Cost 66%

Google launched agentic video understanding in Gemini, letting the model navigate the timeline dynamically instead of ingesting every frame, cutting tokens by up to 88% and costs by 66%. The feature is off by default in the API and requires setting processing to "agentic".

2026-09-07 ~ 2026-09-07 · 2 related posts