fix(utils.py): include system fingerprint in streaming response object

litellm/main.py

@@ -2143,6 +2143,7 @@ def stream_chunk_builder(chunks: list, messages: Optional[list]=None):
     object = chunks[0]["object"]
     created = chunks[0]["created"]
     model = chunks[0]["model"]
+    system_fingerprint = chunks[0].get("system_fingerprint", None)
     role = chunks[0]["choices"][0]["delta"]["role"]
     finish_reason = chunks[-1]["choices"][0]["finish_reason"]
 
@@ -2152,6 +2153,7 @@ def stream_chunk_builder(chunks: list, messages: Optional[list]=None):
         "object": object,
         "created": created,
         "model": model,
+        "system_fingerprint": system_fingerprint,
         "choices": [
             {
                 "index": 0,

litellm/tests/test_stream_chunk_builder.py

@@ -95,16 +95,18 @@ def test_stream_chunk_builder_litellm_tool_call():
     try: 
       litellm.set_verbose = False
       response = litellm.completion(
-          model="azure/chatgpt-functioncalling",
+          model="azure/gpt-4-nov-release",
           messages=messages,
           tools=tools_schema,
           stream=True,
+          api_key="os.environ/AZURE_FRANCE_API_KEY",
+          api_base="https://openai-france-1234.openai.azure.com",
           complete_response = True
       )
 
       print(f"complete response: {response}")
       print(f"complete response usage: {response.usage}")
-
+      assert response.system_fingerprint is not None
     except Exception as e: 
        pytest.fail(f"An exception occurred - {str(e)}")
 

litellm/utils.py

@@ -1796,7 +1796,6 @@ def register_model(model_cost: Union[str, dict]):
         if key in litellm.model_cost:
             for k,v in loaded_model_cost[key].items(): 
                 litellm.model_cost[key][k] = v
-        # litellm.model_cost[key] = loaded_model_cost[key]
         # add new model names to provider lists
         if value.get('litellm_provider') == 'openai':
             if key not in litellm.open_ai_chat_completion_models:
@@ -5291,6 +5290,7 @@ class CustomStreamWrapper:
                 if self.sent_first_chunk == False:
                     model_response.choices[0].delta["role"] = "assistant"
                     self.sent_first_chunk = True
+                # LOGGING
                 threading.Thread(target=self.logging_obj.success_handler, args=(model_response,)).start() # log response
                 return model_response
             else: