From b4ee6ef4f30ae53230a21651055c024c4fd322f3 Mon Sep 17 00:00:00 2001
From: "devin-ai-integration[bot]"
 <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Date: Sun, 25 Oct 2026 02:20:53 +0000
Subject: [PATCH] fix(cost): stamp the billed cost over a reported usage.cost
 on streams

The final streamed usage.cost a client sees is what the proxy billed, so an operator priced deployment shows its own rates instead of the upstream's number. Types region_name on cost_per_token.
---
 litellm/cost_calculator.py                                  | 4 ++--
 litellm/litellm_core_utils/streaming_chunk_builder_utils.py | 2 --
 2 files changed, 2 insertions(+), 4 deletions(-)

diff --git a/litellm/cost_calculator.py b/litellm/cost_calculator.py
index 6ec8d00e37f..cacf130aeab 100644
--- a/litellm/cost_calculator.py
+++ b/litellm/cost_calculator.py
@@ -427,7 +427,7 @@ def cost_per_token(
     completion_tokens: int = 0,
     response_time_ms: float | None = 0.0,
     custom_llm_provider: str | None = None,
-    region_name=None,
+    region_name: str | None = None,
     ### CHARACTER PRICING ###
     prompt_characters: int | None = None,
     completion_characters: int | None = None,
@@ -508,7 +508,7 @@ def _cost_per_token_from_rates(
     completion_tokens: int = 0,
     response_time_ms: float | None = 0.0,
     custom_llm_provider: str | None = None,
-    region_name=None,
+    region_name: str | None = None,
     ### CHARACTER PRICING ###
     prompt_characters: int | None = None,
     completion_characters: int | None = None,
diff --git a/litellm/litellm_core_utils/streaming_chunk_builder_utils.py b/litellm/litellm_core_utils/streaming_chunk_builder_utils.py
index 4cdb9004619..3a5a2cbe914 100644
--- a/litellm/litellm_core_utils/streaming_chunk_builder_utils.py
+++ b/litellm/litellm_core_utils/streaming_chunk_builder_utils.py
@@ -1413,8 +1413,6 @@ def _set_stream_builder_response_cost(
 def _stamp_streaming_usage_cost(usage: Usage, response: ModelResponse, logging_obj: Optional["Logging"]) -> None:
     if logging_obj is None:
         return
-    if isinstance(getattr(usage, "cost", None), (int, float)):
-        return
     computed_cost: Final = logging_obj.response_cost_calculator(result=response)
     if not isinstance(computed_cost, (int, float)) or computed_cost <= 0:
         return