diff --git a/litellm/litellm_core_utils/litellm_logging.py b/litellm/litellm_core_utils/litellm_logging.py
index a59672b5cb2..b5714989cc8 100644
--- a/litellm/litellm_core_utils/litellm_logging.py
+++ b/litellm/litellm_core_utils/litellm_logging.py
@@ -2583,6 +2583,10 @@ class Logging(LiteLLMLoggingBaseClass):
         self._async_success_scheduled = True
         return True
 
+    @property
+    def async_success_claimed(self) -> bool:
+        return self._async_success_scheduled
+
     def mark_logging_complete(
         self,
         event_type: Literal["async_success", "sync_success", "async_failure", "sync_failure"],
diff --git a/litellm/proxy/common_request_processing.py b/litellm/proxy/common_request_processing.py
index 6fda5780f63..6557e5088a3 100644
--- a/litellm/proxy/common_request_processing.py
+++ b/litellm/proxy/common_request_processing.py
@@ -4353,6 +4353,15 @@ class ProxyBaseLLMRequestProcessing:
                     and user_api_key_dict is not None
                 ):
                     await proxy_logging_obj.arelease_max_parallel_requests_on_disconnect(user_api_key_dict)
+            logging_obj: Final = request_data.get("litellm_logging_obj")
+            if (
+                stream_completed
+                and isinstance(logging_obj, LiteLLMLoggingObj)
+                and logging_obj.async_success_claimed
+                and proxy_logging_obj is not None
+                and user_api_key_dict is not None
+            ):
+                await proxy_logging_obj.arelease_max_parallel_requests_on_disconnect(user_api_key_dict)
 
             if hasattr(response, "aclose"):
                 try:
@@ -4362,7 +4371,6 @@ class ProxyBaseLLMRequestProcessing:
                         "async_streaming_data_generator: error closing response stream: %s",
                         e,
                     )
-            logging_obj: Final = request_data.get("litellm_logging_obj")
             if (
                 not stream_completed
                 and isinstance(logging_obj, LiteLLMLoggingObj)