From 34cd139f537434c392d4c491eabb34399d00b96e Mon Sep 17 00:00:00 2001
From: daleselaji-dev <265319989+daleselaji-dev@users.noreply.github.com>
Date: Thu, 27 Aug 2026 12:23:42 +0800
Subject: [PATCH] fix(streaming): avoid rebinding usage helper arguments

---
 .../streaming_chunk_builder_utils.py          | 29 +++++++++++++------
 1 file changed, 20 insertions(+), 9 deletions(-)

diff --git a/litellm/litellm_core_utils/streaming_chunk_builder_utils.py b/litellm/litellm_core_utils/streaming_chunk_builder_utils.py
index 88317817a1e..4c0b06322be 100644
--- a/litellm/litellm_core_utils/streaming_chunk_builder_utils.py
+++ b/litellm/litellm_core_utils/streaming_chunk_builder_utils.py
@@ -762,19 +762,30 @@ class ChunkProcessor:
         prompt_tokens_provided = "prompt_tokens" in usage_chunk
         completion_tokens_provided = "completion_tokens" in usage_chunk
 
-        if prompt_tokens_provided and (usage_chunk_dict["prompt_tokens"] > 0 or prompt_tokens == 0):
-            prompt_tokens = usage_chunk_dict["prompt_tokens"]
-        if completion_tokens_provided and (usage_chunk_dict["completion_tokens"] > 0 or completion_tokens == 0):
-            completion_tokens = usage_chunk_dict["completion_tokens"]
-            if completion_tokens > 0:
-                completion_usage_updates += 1
+        updated_prompt_tokens = (
+            usage_chunk_dict["prompt_tokens"]
+            if prompt_tokens_provided and (usage_chunk_dict["prompt_tokens"] > 0 or prompt_tokens == 0)
+            else prompt_tokens
+        )
+        updated_completion_tokens = (
+            usage_chunk_dict["completion_tokens"]
+            if completion_tokens_provided and (usage_chunk_dict["completion_tokens"] > 0 or completion_tokens == 0)
+            else completion_tokens
+        )
+        updated_completion_usage_updates = (
+            completion_usage_updates + 1
+            if completion_tokens_provided
+            and (usage_chunk_dict["completion_tokens"] > 0 or completion_tokens == 0)
+            and updated_completion_tokens > 0
+            else completion_usage_updates
+        )
 
         return (
-            prompt_tokens,
-            completion_tokens,
+            updated_prompt_tokens,
+            updated_completion_tokens,
             prompt_tokens_provided,
             completion_tokens_provided,
-            completion_usage_updates,
+            updated_completion_usage_updates,
         )
 
     def _calculate_usage_per_chunk(