fix(logging): report the pay-as-you-go entry billing used on Azure PTU spillover (#47567)

get_standard_logging_object_payload decided custom_pricing from
litellm_params alone, so on an Azure PTU spillover response
model_map_information kept the deployment's zeroed PTU entry while
response_cost already used the standard model rate. Reuse the billing
side LitellmLogging._custom_pricing_for so both read the same decision

Backport of #47567 to stable/1.104.x
Cherry-picked from 4222523261 (main)

Co-authored-by: devin-ai-integration[bot] <158243242+devin-ai-integration[bot]@users.noreply.github.com>

litellm/litellm_core_utils/litellm_logging.py

@@ -1882,7 +1882,7 @@ class Logging(LiteLLMLoggingBaseClass):
             router_model_id = self.get_router_model_id()
 
         ## RESPONSE COST ##
-        custom_pricing: Final = self._custom_pricing_for(priced_result)
+        custom_pricing: Final = self.custom_pricing_for(priced_result)
 
         prompt = self._prompt_for_cost_calculation()
 
@@ -2021,7 +2021,7 @@ class Logging(LiteLLMLoggingBaseClass):
             model=model,
             completion_response=result,
             custom_llm_provider=custom_llm_provider,
-            custom_pricing=self._custom_pricing_for(result),
+            custom_pricing=self.custom_pricing_for(result),
             base_model=_get_base_model_from_metadata(model_call_details=self.model_call_details),
             router_model_id=router_model_id or self.get_router_model_id(),
             region_name=_resolve_mantle_region_for_cost(
@@ -2046,7 +2046,7 @@ class Logging(LiteLLMLoggingBaseClass):
             usage=usage,
         )
 
-    def _custom_pricing_for(self, result: object) -> bool:
+    def custom_pricing_for(self, result: object) -> bool:
         litellm_params: Final = getattr(self, "litellm_params", None)
         result_hidden_params: Final = getattr(result, "_hidden_params", None) or MappingProxyType({})
         additional_headers: Final = (
@@ -6525,7 +6525,7 @@ def get_standard_logging_object_payload(
         # Fall back to the actual deployment model set by the router in metadata.
         if base_model is None:
             base_model = metadata.get("deployment")
-        custom_pricing: Final = use_custom_pricing_for_model(litellm_params=litellm_params)
+        custom_pricing: Final = logging_obj.custom_pricing_for(init_response_obj)
         raw_response_cost: Final = kwargs.get("response_cost")
         llm_response_cost: Final[float] = raw_response_cost or 0.0
         guardrail_cost: Final = guardrail_information_cost(metadata.get("standard_logging_guardrail_information"))

tests/unit/litellm_core_utils/test_litellm_logging.py

@@ -27,6 +27,7 @@ from litellm.litellm_core_utils.litellm_logging import Logging as LitellmLogging
 from litellm.litellm_core_utils.litellm_logging import (
     _extract_response_obj_and_hidden_params,
     _get_status_fields,
+    get_standard_logging_object_payload,
     set_callbacks,
 )
 from litellm.llms.base_llm.ocr.transformation import OCRUsageInfo
@@ -8510,6 +8511,40 @@ class TestAzurePTUSpilloverCost:
         finally:
             self._unregister_models()
 
+    def test_spillover_payload_reports_the_pay_as_you_go_entry_billing_used(self, monkeypatch) -> None:
+        self._register_models()
+        try:
+            obj = self._logging_obj(dict(self.PTU_MODEL_INFO), flag="True", litellm_rate=0.0, monkeypatch=monkeypatch)
+            response = self._response()
+            response._hidden_params["additional_headers"] = {"llm_provider-x-ms-is-spilled-over": "true"}
+            response_cost: Final = obj._response_cost_calculator(result=response)
+
+            payload: Final = get_standard_logging_object_payload(
+                kwargs={
+                    "model": self.SERVED_MODEL,
+                    "custom_llm_provider": "azure",
+                    "litellm_params": obj.litellm_params,
+                    "response_cost": response_cost,
+                },
+                init_response_obj=response,
+                start_time=datetime.datetime.now(),
+                end_time=datetime.datetime.now(),
+                logging_obj=obj,
+                status="success",
+            )
+
+            assert response_cost == pytest.approx(self.EXPECTED_SPILL_COST)
+            assert payload is not None
+            model_map_value: Final = payload["model_map_information"]["model_map_value"]
+            assert model_map_value is not None
+            assert model_map_value["input_cost_per_token"] == litellm.model_cost[self.SERVED_MODEL]["input_cost_per_token"]
+            assert model_map_value["output_cost_per_token"] == litellm.model_cost[self.SERVED_MODEL]["output_cost_per_token"]
+            assert response_cost == pytest.approx(
+                100 * model_map_value["input_cost_per_token"] + 50 * model_map_value["output_cost_per_token"]
+            )
+        finally:
+            self._unregister_models()
+
     def test_spillover_header_does_not_touch_non_ptu_custom_pricing(self, monkeypatch) -> None:
         self._register_models()
         custom_model_id: Final = "non-ptu-custom-router-model-id"