Vertex AI AdvancedInstall

Summary

Analyze PDF, image, or text documents using a specified Google Vertex AI model by providing either binary document files or public URLs, along with a prompt or question to apply on the document.

Inputs

• Project ID (required) — The Google Cloud Platform project ID
• Model (required) — The Google Vertex AI model to use for document analysis, chosen from a searchable list or by ID.
• Prompt / Question (required) — Text prompt or question that instructs the model what to analyze or extract from the document.
• Input Type (required) — Specify whether the input document is provided as a binary file or as a URL link.
• Input Data Field Name(s) — The names of binary fields containing the document data when input type is binary; multiple names can be specified separated by commas.
• MIME Type — MIME type of the binary document(s). Choose 'Detect Automatically' to infer from data (recommended).
• URL(s) — Comma-separated URLs of public documents to analyze, used when input type is URL.
• Simplify Output — Whether to simplify the API response to key candidate results or return the full raw response.
• Options — Additional optional settings such as a system message (instruction to the model), maximum output tokens, and parameters controlling output randomness and reasoning token budget.

Output shape

A list of candidate analysis results, each as a JSON object per input item.

If the 'Simplify Output' option is enabled (default), the node returns an array of simplified candidate results from the AI model. Otherwise, it returns the full raw response object from the Vertex AI API. Multiple input documents can be processed either via multiple binary fields or multiple URLs (comma-separated). The node safely extracts and encodes input contents as base64 for the API call.

Links

• n8n community nodes installation guide