Browse all docs

Generate a YouTube thumbnail with AI

PandaStudio generates a YouTube thumbnail from a reaction, a subject image, and overlay text, or suggests one straight from the transcript, using your own AI account.

Describe the thumbnail (an excited reaction, the topic, a few words of overlay) and PandaStudio generates it, or let it read the transcript and propose one. It lives on the export page next to the title and description so you finish everything in one place. It runs OpenAI's GPT Image 2.6 on your own Replicate or Higgsfield account (Settings, Connectors), Replicate first when both are connected: the fast Flare variant for a new thumbnail and Sunburst when you ask for a change to the current one, made at 16:9 so nothing gets cropped off. GPT Image 2.6 needs PandaStudio 2.1.49 or later; earlier versions use GPT Image 2.

In the app

  1. 1Open a finished export's detail page.
  2. 2In the Thumbnail section, pick a reaction and layout, describe the subject, and add overlay text.
  3. 3Generate, or click Suggest from transcript to have it propose one.
  4. 4Set it as the thumbnail before publishing to YouTube.
Export details page, YouTube thumbnail: reaction, layout, subject, overlay text and Suggest from transcript
In the YouTube thumbnail section, describe it or click Suggest from transcript, then Generate thumbnail.

Ask the AI

Open the AI chat in the editor and say any of these. The agent does the edits for you.

“Make a thumbnail with an excited reaction and the words FIRST SALE.”
“Suggest a thumbnail from the transcript.”

CLI & MCP

Every capability is a verb, reachable from the terminal or from an MCP agent (Claude Code, Cursor, Cline, Continue, Claude Desktop).

export.generate-thumbnail
pandastudio export.generate-thumbnail --entryId=$ENTRY --reaction=excited --overlayText='FIRST SALE'
export.set-thumbnail
pandastudio export.set-thumbnail --entryId=$ENTRY --path=<image>