GitHub Copilot Chat API icon

GitHub Copilot Chat API

Use official GitHub Copilot Chat API with your subscription - access GPT-5, Claude, Gemini and more

Chat Completion

AI-generated

Summary

Send a chat message to the GitHub Copilot Chat API using a selected or custom AI model and receive a completion response, optionally including an image in PNG, JPEG, GIF, or WebP format.

Inputs

  • Model (required) — Select the GitHub Copilot model to use, dynamically loaded based on your subscription, or enter a custom model name for new or beta models.
  • Message (required) — The message text to send to the AI model.
  • System Message — Optional system message to define or influence the AI model's behavior.
  • Include Image — Toggle to include an image file with the message for vision-capable models.
  • Image Source — When including an image, choose the source as Manual Input (base64 string or file path), URL (download image), or Binary Data (from a previous node).
  • Image File — Provide the image file as a base64 string or file path (for Manual Input). Supported formats: PNG, JPEG, GIF, WebP.
  • Image URL — Provide the URL to download the image file (for URL source). Supported formats: PNG, JPEG, GIF, WebP.
  • Image Binary Property — Specify the binary property name containing the image file (for Binary Data source). Defaults to 'data'.
  • Advanced Options — Configure additional parameters such as temperature (randomness), max tokens to generate, top_p for diversity, automatic retry on 403 TPM errors, number of retry attempts, and vision fallback to handle images if the primary model lacks vision support.
  • Vision Fallback Model — Select or manually enter a vision-capable fallback model to automatically use when the primary model does not support vision and the message includes images.

Output shape

a list of JSON objects representing individual chat completion responses

Each output item includes the completion message text, the model used, indication if a vision fallback model was utilized, usage statistics if available, finish reason, and retry count when applicable. Supports sending images with models that have vision capabilities or via a configured fallback model. Automatic retry on 403 errors (quota limits) is supported according to configuration.

Links

Discussion