Model Routes

Use Model Routes when your application should call a stable route ID while OpenModels handles the model selection and provider route behind it.

A route ID looks like route:coding-agent. Pass it as the model value in the same POST /chat/completions request shape you already use for normal model IDs.

When to use a Model Route

Use a Model Route when you want to:

• Change routing logic without changing application code.
• Group several models behind one stable ID.
• Fall back across models when a preferred model is unavailable.
• Switch between fast, balanced, and deep modes from your application.

If you only want to pin one request to a provider for a normal model ID, use route.provider on the Chat Completions page instead.

Create a route

Open the OpenModels dashboard, go to Routes, and click Create route.

1. Name the route. OpenModels generates a route ID such as route:coding-agent.
2. Select the models the route can use.
3. Choose a routing rule.
4. Save the route while it is Active.

The route ID cannot be changed after creation. Copy it from the Routes table or the route detail panel.

Priority fallback

Priority fallback uses models in order. If the first model is unavailable, OpenModels falls back to the next model.

{
  "model": "route:coding-agent",
  "route": { "provider": "<provider-for-this-route>" },
  "route_mode": "balanced",
  "messages": [
    {
      "role": "user",
      "content": "Review this patch and summarize the risk."
    }
  ]
}

Mode mapping

Mode mapping lets your application switch between fast, balanced, and deep without changing the route ID. If you omit route_mode, OpenModels uses the balanced mode.

Default balanced mode:

{
  "model": "route:support-chat",
  "route": { "provider": "<provider-for-this-route>" },
  "route_mode": "balanced",
  "messages": [
    {
      "role": "user",
      "content": "Summarize this customer conversation."
    }
  ]
}

Specify a mode:

{
  "model": "route:support-chat",
  "route": { "provider": "<provider-for-this-route>" },
  "route_mode": "deep",
  "messages": [
    {
      "role": "user",
      "content": "Summarize this customer conversation and list follow-up actions."
    }
  ]
}

Supported route_mode values are fast, balanced, and deep.

cURL

curl https://api.getopenmodels.com/v1/chat/completions \
  -H "Authorization: Bearer $OM_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "route:coding-agent",
    "route": { "provider": "<provider-for-this-route>" },
    "route_mode": "balanced",
    "messages": [
      {
        "role": "user",
        "content": "Explain this stack trace in one paragraph."
      }
    ]
  }'

Notes

• Use the route:* ID as the model value.
• Keep using the normal OpenModels base URL and API key.
• Remove route.provider if the route group should select the provider internally.
• Paused routes are not intended for production traffic.
• Usage for each route appears in the dashboard Routes page.