diff --git a/vllm/entrypoints/openai/serving_chat.py b/vllm/entrypoints/openai/serving_chat.py
index 32a3cf04951e..5a916f39b128 100644
--- a/vllm/entrypoints/openai/serving_chat.py
+++ b/vllm/entrypoints/openai/serving_chat.py
@@ -659,9 +659,14 @@ async def chat_completion_stream_generator(
                         "Tokenizer not available when `skip_tokenizer_init=True`"
                     )
 
+                # Pass the same chat template kwargs as used in tokenization
+                chat_template_kwargs = self._prepare_extra_chat_template_kwargs(
+                    request.chat_template_kwargs,
+                    self.default_chat_template_kwargs,
+                )
                 reasoning_parser = self.reasoning_parser(
                     tokenizer,
-                    chat_template_kwargs=request.chat_template_kwargs,  # type: ignore
+                    chat_template_kwargs=chat_template_kwargs,  # type: ignore[call-arg]
                 )
         except RuntimeError as e:
             logger.exception("Error in reasoning parser creation.")
@@ -1437,9 +1442,14 @@ async def chat_completion_full_generator(
                             "Tokenizer not available when `skip_tokenizer_init=True`"
                         )
 
+                    # Pass the same chat template kwargs as used in tokenization
+                    chat_template_kwargs = self._prepare_extra_chat_template_kwargs(
+                        request.chat_template_kwargs,
+                        self.default_chat_template_kwargs,
+                    )
                     reasoning_parser = self.reasoning_parser(
                         tokenizer,
-                        chat_template_kwargs=request.chat_template_kwargs,  # type: ignore
+                        chat_template_kwargs=chat_template_kwargs,  # type: ignore[call-arg]
                     )
                 except RuntimeError as e:
                     logger.exception("Error in reasoning parser creation.")
diff --git a/vllm/entrypoints/openai/serving_engine.py b/vllm/entrypoints/openai/serving_engine.py
index f1c4ab63f05b..e65141edd144 100644
--- a/vllm/entrypoints/openai/serving_engine.py
+++ b/vllm/entrypoints/openai/serving_engine.py
@@ -1146,6 +1146,18 @@ def _validate_chat_template(
             )
         return None
 
+    @staticmethod
+    def _prepare_extra_chat_template_kwargs(
+        request_chat_template_kwargs: dict[str, Any] | None = None,
+        default_chat_template_kwargs: dict[str, Any] | None = None,
+    ) -> dict[str, Any]:
+        """Helper to merge server-default and request-specific chat template kwargs."""
+        request_chat_template_kwargs = request_chat_template_kwargs or {}
+        if default_chat_template_kwargs is None:
+            return request_chat_template_kwargs
+        # Apply server defaults first, then request kwargs override.
+        return default_chat_template_kwargs | request_chat_template_kwargs
+
     async def _preprocess_chat(
         self,
         request: ChatLikeRequest | ResponsesRequest,
@@ -1184,9 +1196,10 @@ async def _preprocess_chat(
             tools=tool_dicts,
             documents=documents,
         )
-        if default_chat_template_kwargs:
-            _chat_template_kwargs.update(default_chat_template_kwargs)
-        _chat_template_kwargs.update(chat_template_kwargs or {})
+        _chat_template_kwargs |= self._prepare_extra_chat_template_kwargs(
+            chat_template_kwargs,
+            default_chat_template_kwargs,
+        )
 
         request_prompt: str | list[int]