Using Your GitHub Copilot Subscription with OpenClaw

If you have a GitHub Copilot subscription, you already have access to some of the most powerful AI models available — Claude, GPT-4o, Gemini, and more. OpenClaw can leverage your existing subscription as an AI provider, giving you a versatile personal assistant without additional API costs.

What You Get

GitHub Copilot's backend provides access to multiple frontier models:

Model	Provider	Best For
claude-opus-4.5	Anthropic	Complex reasoning, long-form content
claude-sonnet-4.5	Anthropic	Balanced speed and capability
gpt-5.2-codex	OpenAI	Code generation and analysis
gpt-4o	OpenAI	General purpose, multimodal
gemini-3-pro-preview	Google	Fast responses, large context

The beauty of this setup: one subscription, multiple models, automatic fallbacks.

Prerequisites

• GitHub Copilot subscription (Individual, Business, or Enterprise)
• OpenClaw installed — Get it from openclaw.ai
• GitHub CLI (gh) — For authentication

Step 1: Install GitHub CLI

If you don't have the GitHub CLI installed:

# macOS
brew install gh

# Ubuntu/Debian
sudo apt install gh

# Windows
winget install GitHub.cli

Step 2: Authenticate with GitHub

Run the GitHub CLI auth flow:

gh auth login

Select:

1. GitHub.com (not Enterprise)
2. HTTPS as the protocol
3. Login with a web browser (or paste a token)

Verify authentication:

gh auth status

You should see your GitHub username and the scopes available.

Step 3: Configure OpenClaw

Run the OpenClaw configuration wizard:

openclaw configure

When prompted for the AI provider, select github-copilot. OpenClaw will automatically detect your GitHub CLI authentication.

Alternatively, configure manually by editing ~/.openclaw/openclaw.json:

{
  "auth": {
    "profiles": {
      "github-copilot:github": {
        "provider": "github-copilot",
        "mode": "token"
      }
    }
  },
  "agents": {
    "defaults": {
      "model": {
        "primary": "github-copilot/claude-opus-4.5",
        "fallbacks": [
          "github-copilot/gpt-5.2-codex",
          "github-copilot/claude-sonnet-4.5",
          "github-copilot/gpt-4o"
        ]
      }
    }
  }
}

Step 4: Choose Your Models

Model Naming Convention

OpenClaw uses the format provider/model-name:

github-copilot/claude-opus-4.5
github-copilot/gpt-5.2-codex
github-copilot/gemini-3-pro-preview

Recommended Configurations

For general assistant use:

"primary": "github-copilot/claude-opus-4.5"

For coding-focused work:

"primary": "github-copilot/gpt-5.2-codex"

For fast responses:

"primary": "github-copilot/claude-haiku-4.5"

Setting Up Fallbacks

Fallbacks activate when the primary model is unavailable or rate-limited:

"model": {
  "primary": "github-copilot/claude-opus-4.5",
  "fallbacks": [
    "github-copilot/gpt-5.2-codex",
    "github-copilot/claude-sonnet-4.5",
    "github-copilot/gemini-3-pro-preview",
    "github-copilot/gpt-4o"
  ]
}

Step 5: Test the Connection

Start the OpenClaw gateway:

openclaw gateway start

Check status:

openclaw status

You should see your configured model and authentication status.

Using Multiple Agents

One powerful feature: you can assign different models to different agents. For example, use a reasoning-heavy model for your main assistant and a fast coding model for subagents:

{
  "agents": {
    "defaults": {
      "model": {
        "primary": "github-copilot/claude-opus-4.5"
      }
    },
    "list": [
      {
        "id": "main"
      },
      {
        "id": "coder",
        "model": "github-copilot/gpt-5.2-codex"
      }
    ]
  }
}

Now your main agent uses Claude for orchestration while delegating coding tasks to GPT Codex.

Checking Usage

OpenClaw tracks your model usage. Check your session stats:

/status

This shows tokens used, model in use, and session time.

Troubleshooting

"Authentication failed"

Re-authenticate with GitHub:

gh auth logout
gh auth login

"Model not available"

Some models may have availability windows or rate limits. Your fallback chain handles this automatically, but you can also try:

openclaw configure --section model

"Token expired"

GitHub CLI tokens refresh automatically, but if you see expiration errors:

gh auth refresh

Cost Considerations

Using GitHub Copilot as your OpenClaw provider means:

• ✅ No additional API costs — Uses your existing subscription
• ✅ Multiple models included — Claude, GPT, Gemini all accessible
• ✅ Automatic rate limit handling — Fallbacks kick in seamlessly
• ⚠️ Subject to Copilot limits — Heavy usage may hit rate limits

For high-volume or production use, consider adding a direct API provider (Anthropic, OpenAI) as a fallback.

Advanced: Mixing Providers

You can combine GitHub Copilot with other providers for maximum flexibility:

{
  "auth": {
    "profiles": {
      "github-copilot:github": {
        "provider": "github-copilot",
        "mode": "token"
      },
      "anthropic:default": {
        "provider": "anthropic",
        "mode": "api_key"
      }
    }
  },
  "agents": {
    "defaults": {
      "model": {
        "primary": "github-copilot/claude-opus-4.5",
        "fallbacks": [
          "anthropic/claude-sonnet-4-20250514",
          "github-copilot/gpt-4o"
        ]
      }
    }
  }
}

This setup uses your Copilot subscription first, falls back to direct Anthropic API if needed.

Conclusion

Your GitHub Copilot subscription is more than just code completion — it's a gateway to frontier AI models. By configuring OpenClaw to use it as a provider, you get:

• Access to Claude, GPT, and Gemini models
• No additional API subscriptions required
• Automatic model fallbacks
• The full power of OpenClaw's agent system

If you're already paying for Copilot, you're leaving capability on the table by not using it with OpenClaw.

---

For more configuration options, see the OpenClaw documentation. Join the community on Discord.