fix ResponseAPILoggingUtils

litellm/cost_calculator.py

@@ -45,7 +45,11 @@ from litellm.llms.vertex_ai.image_generation.cost_calculator import (
     cost_calculator as vertex_ai_image_cost_calculator,
 )
 from litellm.responses.utils import ResponseAPILoggingUtils
-from litellm.types.llms.openai import HttpxBinaryResponseContent, ResponsesAPIResponse
+from litellm.types.llms.openai import (
+    HttpxBinaryResponseContent,
+    ResponseAPIUsage,
+    ResponsesAPIResponse,
+)
 from litellm.types.rerank import RerankBilledUnits, RerankResponse
 from litellm.types.utils import (
     CallTypesLiteral,
@@ -465,6 +469,13 @@ def _get_usage_object(
     return usage_obj
 
 
+def _is_known_usage_objects(usage_obj):
+    """Returns True if the usage obj is a known Usage type"""
+    return isinstance(usage_obj, litellm.Usage) or isinstance(
+        usage_obj, ResponseAPIUsage
+    )
+
+
 def _infer_call_type(
     call_type: Optional[CallTypesLiteral], completion_response: Any
 ) -> Optional[CallTypesLiteral]:
@@ -588,8 +599,8 @@ def completion_cost(  # noqa: PLR0915
                 )
             else:
                 usage_obj = getattr(completion_response, "usage", {})
-            if isinstance(usage_obj, BaseModel) and not isinstance(
-                usage_obj, litellm.Usage
+            if isinstance(usage_obj, BaseModel) and not _is_known_usage_objects(
+                usage_obj=usage_obj
             ):
                 setattr(
                     completion_response,

tests/llm_responses_api_testing/test_openai_responses_api.py

@@ -13,6 +13,8 @@ from litellm.types.llms.openai import (
     ResponseCompletedEvent,
     ResponsesAPIResponse,
     ResponseTextConfig,
+    ResponseAPIUsage,
+    IncompleteDetails,
 )
 
 
@@ -50,7 +52,7 @@ def validate_responses_api_response(response):
     # Optional fields with their expected types
     optional_fields = {
         "error": (dict, type(None)),  # error can be dict or None
-        "incomplete_details": (dict, type(None)),
+        "incomplete_details": (IncompleteDetails, type(None)),
         "instructions": (str, type(None)),
         "metadata": dict,
         "model": str,
@@ -65,7 +67,7 @@ def validate_responses_api_response(response):
         "status": str,
         "text": ResponseTextConfig,
         "truncation": str,
-        # "usage": dict,
+        "usage": ResponseAPIUsage,
         "user": (str, type(None)),
     }
 
@@ -89,9 +91,13 @@ async def test_basic_openai_responses_api(sync_mode):
     litellm._turn_on_debug()
 
     if sync_mode:
-        response = litellm.responses(model="gpt-4o", input="Basic ping")
+        response = litellm.responses(
+            model="gpt-4o", input="Basic ping", max_output_tokens=20
+        )
     else:
-        response = await litellm.aresponses(model="gpt-4o", input="Basic ping")
+        response = await litellm.aresponses(
+            model="gpt-4o", input="Basic ping", max_output_tokens=20
+        )
 
     print("litellm response=", json.dumps(response, indent=4, default=str))