Google Vertex AI icon

Google Vertex AI

Use Google Vertex AI (Gemini) for text, media analysis, and image generation/editing

Actions11

Image → Analyze Image

AI-generated

Summary

Sends one or more images from the incoming item's binary properties, plus a text prompt, to a Gemini model on Vertex AI and returns the model's answer about the image(s).

Inputs

  • Images (required) — List of images to analyze. Each entry gives the name of a binary property on the incoming item (default "data"). Names are trimmed and blank ones are ignored. At least one is required, or the node errors. The images are base64-encoded inline and sent using the binary's own MIME type. Entries can be reordered.
  • Prompt (required) — The question or instruction about the image(s). It is sent as text after the image parts.
  • Model (required) — Gemini model, chosen from a list or entered by ID (default gemini-2.5-flash). This comes from the node's shared fields rather than the operation-scoped properties.
  • Options — Shared collection: System Instruction, Temperature, Max Output Tokens, Top P, Top K, JSON Output (requests application/json), and Simplify Output (default on).

Output shape

One output item per input item. With Simplify Output on (the default), the JSON is {"text": ""}. With it off, the JSON is the full Vertex AI generateContent response.

The text is the concatenated text parts of the first candidate. Each result is paired to its input item. If Continue On Fail is enabled, a failed item returns {"error": message}. Requests use the Google Vertex AI service-account credential (project and region come from the credential). Missing binary data or API errors raise a node error, and some errors (401/403/404/429) get a friendlier hint.

Examples

Example 1: Describe an image downloaded by an HTTP Request node.

Images: one entry with binary field "data"; Prompt: "Describe this image in one paragraph."; default model and options.

Example 2: Compare two uploaded product photos and get structured JSON.

Images: two entries ("before", "after"); Prompt: "List the visible differences as a JSON array."; Options: JSON Output on.

Links

Discussion