Fix(dev): catch llm configuration error on get suggestions (#12647)

src/documents/tests/test_views.py

@@ -356,6 +356,31 @@ class TestAISuggestions(DirectoriesMixin, TestCase):
             },
         )
 
+    @patch("documents.views.get_ai_document_classification")
+    @override_settings(
+        AI_ENABLED=True,
+        LLM_BACKEND="openai",
+    )
+    def test_suggestions_with_invalid_ai_configuration(
+        self,
+        mock_get_ai_classification,
+    ) -> None:
+        mock_get_ai_classification.side_effect = ValueError(
+            "Unknown model 'gpt-5.4-mini-2026-03-17'.",
+        )
+
+        self.client.force_login(user=self.user)
+        response = self.client.get(f"/api/documents/{self.document.pk}/suggestions/")
+
+        self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+        self.assertEqual(
+            response.json(),
+            {
+                "ai": ["Invalid AI configuration."],
+            },
+        )
+        self.assertIsNone(get_llm_suggestion_cache(self.document.pk, backend="openai"))
+
     def test_invalidate_suggestions_cache(self) -> None:
         self.client.force_login(user=self.user)
         suggestions = {