fix(quotas): list unpriced models by recorded cost; integer token limits in status

The unpriced-model notice asked the live registry whether a model has a
price, so a priced model whose provider was later disabled showed up as
unpriced. It now lists models whose calls this period were all recorded at
$0. Token limits are serialized as integers.

docsgpt/api/admin/quotas.py

@@ -151,13 +151,15 @@ def _delete_policy(scope: str, subject_id: Optional[str]):
 
 
 def _unpriced_models(conn) -> list[dict]:
-    """Catalog models used this period that no cost limit can see."""
+    """Models used this period whose calls were all recorded at $0 for want of a price."""
     start, _ = window_bounds(settings.QUOTA_PERIOD)
     return [
         row
         for row in TokenUsageRepository(conn).tokens_by_model(start=start)
-        # BYOM ids are UUIDs; those calls are $0 by design, not by omission.
-        if not looks_like_uuid(row["model_id"]) and not is_priced(row["model_id"])
+        # Judged by what was recorded, so a priced model whose provider has since
+        # been disabled is not listed. BYOM ids are UUIDs and $0 by design; a
+        # model explicitly priced at $0 is free, not unpriced.
+        if row["cost"] == 0 and not looks_like_uuid(row["model_id"]) and not is_priced(row["model_id"])
     ]
 
 

docsgpt/quotas/service.py

@@ -43,18 +43,19 @@ class BucketStatus:
     def to_dict(self) -> dict:
         """Return the JSON shape shared by the admin and user quota endpoints."""
 
-        def budget(limit: ResolvedLimit, used: float) -> dict:
+        def budget(limit: Optional[float], resolved: ResolvedLimit, used: float) -> dict:
             return {
-                "limit": limit.limit,
+                "limit": limit,
                 "used": used,
-                "source": limit.source,
-                "source_id": limit.source_id,
+                "source": resolved.source,
+                "source_id": resolved.source_id,
             }
 
+        tokens, cost = self.limits.tokens, self.limits.cost
         return {
             "bucket": self.bucket,
-            "tokens": budget(self.limits.tokens, self.tokens_used),
-            "cost": budget(self.limits.cost, round(self.cost_used, 6)),
+            "tokens": budget(None if tokens.unlimited else int(tokens.limit), tokens, self.tokens_used),
+            "cost": budget(cost.limit, cost, round(self.cost_used, 6)),
             "resets_at": self.resets_at.isoformat(),
         }