OpenClaw

Before you start

1. Create an API key. In the portal, open Keys and click Create Key. Your key starts with ni_live_.
2. Add credit. New accounts start with a balance of $0, and every request is refused until you add credit. Open Billing and click Buy credits.
3. Pick a model. Northern Inference model names end in a tier tag, for example claude-sonnet-4-6-t4. See them all on the Models page.

Install OpenClaw

Mac, Linux and WSL2:

curl -fsSL https://openclaw.ai/install.sh | bash

Windows PowerShell:

iwr -useb https://openclaw.ai/install.ps1 | iex

On Windows you can also install the OpenClaw desktop app from openclaw.ai. The installer starts OpenClaw's setup (onboarding) when it finishes.

Connect OpenClaw to Northern Inference

When onboarding asks for a model provider, choose the custom OpenAI-compatible endpoint and enter:

To do the same from the command line, run:

openclaw onboard --auth-choice custom-api-key --custom-base-url https://northerninference.ca/v1 --custom-model-id claude-sonnet-4-6-t4 --custom-compatibility openai --custom-api-key ni_live_YOUR_KEY_HERE

Send a message and check that the request appears in the portal under Usage within a few seconds.

Or edit the config file

If you prefer to write the configuration yourself, put your key in ~/.openclaw/.env (on Windows, C:\Users\you\.openclaw\.env):

NI_API_KEY=ni_live_YOUR_KEY_HERE

Then put this in ~/.openclaw/openclaw.json (on Windows, C:\Users\you\.openclaw\openclaw.json). If the file already exists, add the northerninference provider and the agents default to what is there instead of replacing the file.

{
  models: {
    providers: {
      northerninference: {
        baseUrl: "https://northerninference.ca/v1",
        apiKey: "${NI_API_KEY}",
        api: "openai-completions",
        models: [
          {
            id: "claude-sonnet-4-6-t4",
            name: "Claude Sonnet 4.6 (Northern Inference)",
            input: ["text"],
            contextWindow: 128000,
            maxTokens: 16384,
            // USD per 1M tokens, from https://northerninference.ca/api/billing/models
            cost: { input: 3, output: 15, cacheRead: 0.3, cacheWrite: 3.75 },
          },
        ],
      },
    },
  },
  agents: {
    defaults: {
      model: { primary: "northerninference/claude-sonnet-4-6-t4" },
    },
  },
}

To add another model, add another entry to models with its exact name from the Models page, for example DeepSeek-V4-Pro-t4.

Show token counts and cost in OpenClaw (optional)

Northern Inference bills the exact provider counts either way, and the portal always shows them. To see them inside OpenClaw as well:

1. In the portal, open your API key and turn on Always send streaming token counts. OpenClaw streams every request but does not ask for the token counts, so Northern Inference only sends them when this is on.
2. Keep the cost block on each model entry. OpenClaw works out the cost itself from it, in US dollars per million tokens. The numbers above change when providers change their prices; the current ones are on the Models page.

If a model name is not recognized

Northern Inference answers with HTTP 404 and Model 'X' is not a route this key can call. It never swaps in a different model. Use the exact name from the Models page, including its case, or type ni-models as a chat message to list the models your key can call.

Troubleshooting