OpenClaw on Mingles Router

Run OpenClaw on Mingles Router

Add Mingles Router as a custom OpenAI-compatible provider in OpenClaw and route any agent to Kimi-K2.6, Qwen3 or MiniMax.

Get a free API key OpenClaw on GitHub

Pick your model and paste your key — every command and value on this page updates live.

Endpoint: https://router.mingles.ai/v1 · key stays in your browser · no key entered? commands show <your-key>.

Install + configure

1. Install OpenClaw

Install the CLI and run onboarding once.

shell
npm install -g openclaw
openclaw onboard

2. Add Mingles Router as a custom provider

OpenClaw reads ~/.openclaw/openclaw.json. Add a provider under models.providers with the openai-completions adapter:

~/.openclaw/openclaw.json
{
  "models": {
    "providers": {
      "router": {
        "baseUrl": "https://router.mingles.ai/v1",
        "apiKey": "<your-key>",
        "api": "openai-completions",
        "models": [
          {
            "id": "moonshotai/Kimi-K2.6",
            "name": "moonshotai/Kimi-K2.6",
            "contextWindow": 200000,
            "maxTokens": 8192,
            "input": ["text"]
          }
        ]
      }
    }
  }
}

If api is omitted on a provider that has a baseUrl, OpenClaw defaults to openai-completions anyway.

3. Allow the model (required)

OpenClaw rejects any model that is not on the agent allowlist. Add it under agents.defaults.models and set it as primary:

~/.openclaw/openclaw.json
{
  "agents": {
    "defaults": {
      "model": { "primary": "router/moonshotai/Kimi-K2.6" },
      "models": {
        "router/moonshotai/Kimi-K2.6": { "alias": "Mingles Router" }
      }
    }
  }
}

Do not edit ~/.openclaw/agents/main/agent/models.json directly — it is regenerated from openclaw.json.

4. Run it

Pick the model and confirm it is live:

shell
openclaw models set router/moonshotai/Kimi-K2.6
openclaw models list
openclaw

Exact values

Config file ~/.openclaw/openclaw.json
Provider key router
baseUrl https://router.mingles.ai/v1
apiKey <your-key>
api (adapter) openai-completions
Model ref router/moonshotai/Kimi-K2.6

Switch model

All three models live on the same provider, so switching is one command — just change the model id (no new key):

shell
openclaw models set router/moonshotai/Kimi-K2.6

Before you file a bug: read the limits

Reasoning models spend the output budget on internal thinking, output is capped at 8192 tokens, there is no KV cache, and there are no built-in web tools. Most “it broke on Mingles Router” reports are one of these. See Model limits & behavior →

Frequently asked

Why does OpenClaw reject my model? +

A provider definition alone is not enough — the model id must also be on the agents.defaults.models allowlist. Add the provider/model entry and set model.primary to it.

What adapter value do I use? +

api: "openai-completions" — the OpenAI /v1/chat/completions dialect. For non-native endpoints OpenClaw also disables the developer role automatically to avoid 400s.

How is the model referenced? +

As provider/model — e.g. router/moonshotai/Kimi-K2.6. Use that exact string in model.primary and openclaw models set.

Where does the API key go? +

Into apiKey on the provider in ~/.openclaw/openclaw.json (or an ${ENV_VAR} reference). It never leaves your machine.

Get API key More agent runtimes