Corion
Get started

Using Corion AI with OpenClaw

OpenClaw connects to Corion either through its onboarding wizard or a manual models.providers entry in ~/.openclaw/openclaw.json.

Prerequisites

Wizard path (recommended)

openclaw configure        # or `openclaw onboard` on a fresh install

Choose Local (this machine) for the Gateway, go to the Model section, and pick Custom provider. It walks you through:

  • API Base URL — use Corion's root endpoint:

    https://api.corion.ai/v1

  • Endpoint compatibility — openai-completions (OpenAI-style /v1/chat/completions). This is the right choice for Corion; openai-responses, anthropic-messages, and Unknown (auto-detect) are not needed.

  • API key — your Corion key. Stored as an auth profile (recommended location: ~/.openclaw/.env or the Gateway process environment).

  • Model ID — the exact Corion model name: kimi-k3, deepseek-v4-pro-0813 or deepseek-v4-flash-0731.

  • Alias — optional friendly name (e.g. k3).

  • Endpoint ID — short id that becomes the provider half of your model ref, e.g. corion.

Adding auth does not change your primary model. To switch, run:

openclaw models set corion/kimi-k3

Manual config

Edit ~/.openclaw/openclaw.json and add the provider under models.providers:

{
  "models": {
    "mode": "merge",
    "providers": {
      "corion": {
        "baseUrl": "https://api.corion.ai/v1",
        "apiKey": "your-corion-api-key",
        "api": "openai-completions",
        "models": [
          {
            "id": "kimi-k3",
            "reasoning": true,
            "input": ["text", "image"],
            "contextWindow": 1048576,
            "maxTokens": 131072,
            "cost": { "input": 2.5, "output": 12.5, "cacheRead": 0.25, "cacheWrite": 0 }
          },
          {
            "id": "deepseek-v4-pro-0813",
            "reasoning": true,
            "input": ["text"],
            "contextWindow": 1048576,
            "maxTokens": 384000,
            "cost": { "input": 0.435, "output": 0.87, "cacheRead": 0.0435, "cacheWrite": 0 }
          },
          {
            "id": "deepseek-v4-flash-0731",
            "reasoning": true,
            "input": ["text"],
            "contextWindow": 1048576,
            "maxTokens": 65536,
            "cost": { "input": 0.11, "output": 0.25, "cacheRead": 0.011, "cacheWrite": 0 }
          }
        ]
      }
    }
  },
  "agents": {
    "defaults": {
      "model": { "primary": "corion/kimi-k3" }
    }
  }
}

Then apply (the Gateway restarts to pick it up):

openclaw gateway config.apply --file ~/.openclaw/openclaw.json

Verify

openclaw models

Confirm corion/kimi-k3, corion/deepseek-v4-pro-0813 and corion/deepseek-v4-flash-0731 appear. You can also confirm the key sees the upstream list directly:

curl https://api.corion.ai/v1/models -H "Authorization: Bearer your-corion-api-key"

Run a read-only job first ("summarize this directory") and confirm both the reply and tool calls before letting OpenClaw write code. If a request returns 401, the key is wrong or incomplete — Corion shows the complete key only once at creation, so generate a fresh one in the dashboard. For other errors, see errors and retries.