> ## Documentation Index
> Fetch the complete documentation index at: https://docs.belvedir.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# POST /api/v1/route: Ask Which Model Should Serve a Call

> Ask Belvedir which model should serve a call, decided from your project's routing config and task groups. Deterministic and fast (no model call).

Asks which model should serve a call, decided from the project's Routing config (default model, optional strong model, per-task-group overrides) plus a lexical match against your task groups and a complexity score on the request. Deterministic and fast; no model call is made. Same API-key auth as ingest.

* **Authorization**: Bearer token with your API key, or use the `x-api-key` header.
* **Body**: `{ "messages": [...] }` or `{ "task": "summarize this doc" }`

## Response

```json theme={null}
{
  "configured": true,
  "model": "qwen3:4b",
  "tier": "default",
  "matched_group": null,
  "complexity": 0.35,
  "reason": "..."
}
```

`tier` is `"group"` when a pinned task group matched, `"strong"` when the complexity score crossed the escalation threshold, else `"default"`. An unconfigured project gets `configured: false` (never an error), so the call is safe to ship before the dashboard config exists; keep a local fallback model.

See [Model Routing](/inference/mixture-of-models) for the full integration pattern.