From 804511571970984fe80577129ee78f1c4877f2c6 Mon Sep 17 00:00:00 2001
From: PRABHU KIRAN VANDRANKI <72809214+VANDRANKI@users.noreply.github.com>
Date: Mon, 8 Jun 2026 13:14:13 -0400
Subject: [PATCH] fix(proxy): forward proxy headers to batches API via
 extra_headers

Continuation of the files-endpoint fix for #27641.

Apply the same `data["headers"] -> data["extra_headers"]` remap to all four
batch handlers (create, retrieve, list, cancel) so that headers injected by
`add_user_information_to_llm_headers` and `forward_client_headers_to_llm_api`
reach the upstream provider.
---
 litellm/proxy/batches_endpoints/endpoints.py | 28 ++++++++++++++++++++
 1 file changed, 28 insertions(+)

diff --git a/litellm/proxy/batches_endpoints/endpoints.py b/litellm/proxy/batches_endpoints/endpoints.py
index 935b96a0e39..781ef28f5f6 100644
--- a/litellm/proxy/batches_endpoints/endpoints.py
+++ b/litellm/proxy/batches_endpoints/endpoints.py
@@ -107,6 +107,13 @@ async def create_batch(  # noqa: PLR0915
             route_type="acreate_batch",
         )
 
+        # Remap proxy-injected headers to extra_headers;
+        # batch TypedDicts use extra_headers, not headers (fixes #27641)
+        if "headers" in data:
+            _proxy_headers = data.pop("headers")
+            if _proxy_headers:
+                data["extra_headers"] = {**(data.get("extra_headers") or {}), **_proxy_headers}
+
         ## check if model is a loadbalanced model
         router_model: Optional[str] = None
         is_router_model = False
@@ -394,6 +401,13 @@ async def retrieve_batch(  # noqa: PLR0915
             route_type="aretrieve_batch",
         )
 
+        # Remap proxy-injected headers to extra_headers;
+        # batch TypedDicts use extra_headers, not headers (fixes #27641)
+        if "headers" in data:
+            _proxy_headers = data.pop("headers")
+            if _proxy_headers:
+                data["extra_headers"] = {**(data.get("extra_headers") or {}), **_proxy_headers}
+
         # FIX: First, try to read from ManagedObjectTable for consistent state
         managed_files_obj = proxy_logging_obj.get_proxy_hook("managed_files")
         from litellm.proxy.proxy_server import prisma_client
@@ -653,6 +667,13 @@ async def list_batches(
             route_type="alist_batches",
         )
 
+        # Remap proxy-injected headers to extra_headers;
+        # batch TypedDicts use extra_headers, not headers (fixes #27641)
+        if "headers" in data:
+            _proxy_headers = data.pop("headers")
+            if _proxy_headers:
+                data["extra_headers"] = {**(data.get("extra_headers") or {}), **_proxy_headers}
+
         # Try to use managed objects table for listing batches (returns encoded IDs)
         managed_files_obj = proxy_logging_obj.get_proxy_hook("managed_files")
         if managed_files_obj is not None and hasattr(
@@ -852,6 +873,13 @@ async def cancel_batch(
             proxy_config=proxy_config,
         )
 
+        # Remap proxy-injected headers to extra_headers;
+        # batch TypedDicts use extra_headers, not headers (fixes #27641)
+        if "headers" in data:
+            _proxy_headers = data.pop("headers")
+            if _proxy_headers:
+                data["extra_headers"] = {**(data.get("extra_headers") or {}), **_proxy_headers}
+
         # SCENARIO 1: Batch ID is encoded with model info
         if model_from_id is not None:
             credentials = get_credentials_for_model(