Analyzing PDFs and Videos with Gemini in Claude Code Using the Gemini API CLI

At a glance

Google's newly released Gemini API CLI lets coding tools such as Claude Code and Codex hand off PDFs, images, and videos to Gemini for analysis from the terminal. It runs as a command-line tool (a program you operate by typing text commands in a terminal). You only need an API key from Google AI Studio, and it can be used on the free tier.

Key points

1. 1

   Using the analyze command, you can ask questions about PDFs, spreadsheets, images, audio, videos, and public YouTube links and get answers.

2. 2

   Large files must be uploaded first, and then you include the returned address in the analysis. Local files up to 14MB in original size are included directly.

3. 3

   Using the agent run command, you can run deep research (a feature that searches the web many times, checks sources, and writes a report) in the background and check its status later.

4. 4

   Korea is included among the supported regions. The main model used for analysis is free for input and output on the free tier, but paid tiers will see price increases starting January 1, 2027.

Why it matters

Instead of reading long documents or meeting recordings yourself, you can have an AI tool handle the analysis, which is drawing attention as a way to save work time.