Skip to main content
Edgee’s OpenAI-compatible API works with OpenClaw, letting you route all model traffic through Edgee to gain token compression, intelligent routing, and observability — without changing how you use OpenClaw day-to-day.

​
Setup

1

Get your Edgee API key

In the Edgee Console, navigate to API Keys, click Create API Key, and copy the key. It will not be shown again.
No Edgee account yet? Create one here — the free tier includes observability and compression with no credit card required.
2

Add Edgee as a provider in OpenClaw

Open your OpenClaw config file at ~/.openclaw/openclaw.json (create it if it doesn’t exist) and add the edgee provider under models.providers:
{
  models: {
    mode: "merge",
    providers: {
      edgee: {
        baseUrl: "https://api.edgee.ai/v1",
        apiKey: "sk-edgee-...",   // replace with your key
        api: "openai-completions",
        models: [
          {
            id: "claude-sonnet-4-6",
            name: "Claude Sonnet 4.6 (Edgee)",
            input: ["text"],
            contextWindow: 200000,
            maxTokens: 8096
          },
          {
            id: "gpt-4o",
            name: "GPT-4o (Edgee)",
            input: ["text", "image"],
            contextWindow: 128000,
            maxTokens: 16384
          }
        ]
      }
    }
  }
}
Replace "sk-edgee-..." with your Edgee API key. Add or remove models from the list to match what you want to use — see the Edgee model list for all available model IDs.
You can also set the key via an environment variable to avoid storing it in the file:
apiKey: "${EDGEE_API_KEY}"
Then export EDGEE_API_KEY in your shell profile.
3

Select an Edgee model in OpenClaw

Restart OpenClaw. Your Edgee models will appear in the model picker under the names you defined (e.g., “Claude Sonnet 4.6 (Edgee)”).Select one to start routing requests through Edgee.

​
Benefits

Token compression

Edgee trims tool results, reduces tool surface area, and optionally shortens model output — cutting token costs with no change to output quality.

Any model, one key

Access models from Anthropic, OpenAI, Mistral, and more through a single Edgee API key — no per-provider account management.

Automatic fallback

If a provider returns a 5xx or times out, Edgee reroutes to the next available provider. Your OpenClaw session continues uninterrupted.

Observability

Every request is logged in the Edgee Console with token counts, latency, compression savings, and cost breakdowns — per session and across your team.
Token compression applies to all requests routed through Edgee. You can bring your own provider keys via BYOK — compression still applies as long as requests pass through Edgee.

​
Next Steps