Get Used Models

Return the distinct models the user has run inference against.

One entry per distinct inferences.model_id for the authenticated user, ordered by most-recently-used first and carrying the last-used timestamp plus a lifetime call count. LLM-as-Judge calls (source = 'llmaj') are excluded so the list reflects user-driven usage. The service filters inferences.user_id explicitly because the service_role async engine bypasses Postgres RLS.

Args: limit: Maximum number of models to return. auth: Authentication result carrying the user id used to scope the aggregate.

Returns: UsedModelsResponse with the recency-ordered model list. Empty when the user has no qualifying inferences.

Raises: HTTPException: 503 when the aggregate fails; the cause is logged server-side rather than returned to the caller.