GitHub Copilot can be used as an AI provider through the OpenClaw gateway, routing requests through your organisation’s Copilot subscription.
Prerequisites
- A GitHub account with an active GitHub Copilot subscription (Individual, Business, or Enterprise)
- The
openclaw CLI version 1.3.0 or later
Authentication
Use the interactive device-flow login:
openclaw models auth login-github-copilot
| Variable | Notes |
|---|
COPILOT_GITHUB_TOKEN | Explicit Copilot token — highest priority |
GH_TOKEN | GitHub CLI token — Copilot activates automatically if this is already exported |
GITHUB_TOKEN | Generic GitHub token — lowest-priority fallback |
GitHub Copilot tokens expire. Run openclaw models auth login-github-copilot again if you see authentication errors.
Configuration
# No config file entry needed — Copilot is activated via env var:
# COPILOT_GITHUB_TOKEN=<your-token>
# Or run: openclaw models auth login-github-copilot
The API endpoint (https://api.individual.githubcopilot.com) is also dynamically derived from the Copilot token exchange — do not change this unless using a proxy.
Available models
Models vary by subscription tier:
| Model ID | Notes |
|---|
gpt-4o | Default for Business/Enterprise |
gpt-4.1 | |
gpt-4.1-mini | |
gpt-4.1-nano | |
o3-mini | Reasoning model |
o1 | Reasoning model |
o1-mini | Reasoning model |
claude-sonnet-4.6 | Available on Enterprise |
claude-sonnet-4.5 | Available on Enterprise |
openclaw models status --provider copilot to see the current list for your subscription.
Verify the connection
openclaw models status --probe-provider github-copilot
