OpenClaw + TOKNADO

How to use OpenClaw with TOKNADO's OpenAI-compatible endpoint

OpenClaw lets you plug in any endpoint that speaks the OpenAI chat completions format as a custom provider. TOKNADO is exactly that: one OpenAI-compatible endpoint in front of a managed model pool, so OpenClaw gets a single base URL and key instead of several provider accounts.

This guide covers both ways to add it: the interactive wizard (fastest) and the manual config for people who keep their setup in version control. It also covers the one step people miss: allowing the model after you define the provider.

Start free trial Jump to setup

What you need

• A TOKNADO API key and endpoint from your dashboard (the trial is enough to test).
• OpenClaw installed and able to run openclaw configure.
• A model ID to use. Copy it exactly as it appears in your TOKNADO dashboard.

Setup: OpenClaw with TOKNADO

1. Run the configuration wizard

   On an existing install this adds a provider without disturbing what you already have.

   openclaw configure

2. Choose the custom provider path

   When asked where the Gateway will run, choose Local. For the section, choose Model. For Model / Auth Provider, choose Custom provider.

3. Enter the TOKNADO details

   Use these values at the prompts. For compatibility, pick OpenAI-compatible. The base URL must include the API path (normally ending in /v1).

   API Base URL  : https://toknado.up.railway.app/v1
   API key       : YOUR_TOKNADO_API_KEY
   Compatibility : OpenAI-compatible
   Model ID      : toknado/auto

4. Or edit the config directly

   If you prefer manual config, add a provider entry. api must be openai-completions for OpenAI-compatible endpoints.

   {
     "models": {
       "mode": "merge",
       "providers": {
         "toknado": {
           "baseUrl": "https://toknado.up.railway.app/v1",
           "apiKey": "YOUR_TOKNADO_API_KEY",
           "api": "openai-completions",
           "models": [
             { "id": "toknado/auto", "name": "TOKNADO" }
           ]
         }
       }
     }
   }

5. Select the model and test

   Model references use the form provider/model-id. Set it as default, then open the dashboard and send a message.

   openclaw models set toknado/toknado/auto
   openclaw dashboard

Notes for real-world use

• Agent tools make many model calls per task. TOKNADO limits requests over a rolling 60 minutes, so a busy OpenClaw gateway on a small plan is the first place you would notice the cap. The plan table below lists each hourly limit.
• “OpenAI-compatible” guarantees the chat completions shape, not every optional feature. If tool calls or reasoning controls misbehave, test with a plain prompt first, then add features back one at a time.
• Keep the API key out of version control. Use an environment variable or your secrets manager and reference it from the config where OpenClaw supports that.

Troubleshooting OpenClaw + TOKNADO

FAQ