Skip to content
Databraid Docs

Anthropic Vision Node

The Anthropic Vision Node allows you to analyze images using Anthropic’s Claude models. You can provide one or multiple images, either as URLs or base64-encoded data, along with an optional system prompt and a user prompt. The node will send the images and prompts to the Anthropic API and return the generated analysis.

Inputs

  • images: An array of image URLs or base64-encoded image data.
  • sys_prompt: (Optional) A string representing the system prompt to guide the analysis.
  • prompt: A string representing the user prompt or question for the image analysis.

Outputs

  • output: The generated analysis from the Anthropic API.

Properties

  • model: The Anthropic model to use for image analysis (default: “claude-3-opus-20240229”).
  • secret: The API key for accessing Anthropic services.

Usage

  1. Connect an array of image URLs or base64-encoded image data to the “images” input.
  2. (Optional) Connect a string to the “sys_prompt” input to provide a system prompt.
  3. Connect a string to the “prompt” input to provide a user prompt or question for the image analysis.
  4. Set the “secret” property to your Anthropic API key.
  5. The node will process the images and prompts and output the generated analysis.

Tips

  • Make sure to provide valid image URLs or base64-encoded image data.
  • If an image URL is invalid or cannot be fetched, an error will be logged in the console, and the image will be excluded from the analysis.
  • Adjust the “model” property if you want to use a different Anthropic model for image analysis.