Skip to main content
After the connector starts and establishes its tunnel, you register the deployment in the Tokios dashboard. Registration gives your model a stable name you use as the model field in every API call — for example, gemma-tunnel or llama3-local.

​
Register Your Deployment

1

Open the dashboard

Go to your Tokios account dashboard. When the connector is running and the tunnel is live, you’ll see its status displayed as Connected.If the status shows Waiting or Disconnected, check that the connector is running on your local machine and that tokios.json contains the correct tunnel_token.
2

Pair the connector

Click Pair Connector. The dashboard will either:
  • Auto-detect the running connector and prompt you to confirm the pairing, or
  • Ask you to enter the pairing code shown in the connector’s terminal output.
The pairing code is displayed in the connector’s startup logs, directly below the Tunnel established message.
3

Name the deployment

Choose a model name for this deployment — for example, gemma-tunnel or llama3-local. This is the exact string you will pass as the "model" field in every API request routed to this connector.Naming rules:
  • Lowercase alphanumeric characters only
  • Hyphens (-) are allowed
  • No spaces, underscores, or special characters
Good examples: gemma-tunnel, llama3-local, mistral-home
4

Save

Click Register. The model now appears as an active deployment under your account and is immediately reachable at api.tokios.com.

​
Using Your Model Name in API Calls

Once registered, pass the deployment name you chose as the model field in any API request: 
{
  "model": "gemma-tunnel",
  "messages": [
    { "role": "user", "content": "What is the capital of France?" }
  ]
}
The Tokios gateway resolves gemma-tunnel to the paired connector and forwards the request to your local model server in real time.
You can register multiple models by running multiple connectors with different upstream ports, each paired separately in the dashboard. For example, one connector pointing at Ollama on port 11434 and another pointing at vLLM on port 8000 can both be active and registered at the same time.