diff --git a/python/sglang/srt/openai_api/adapter.py b/python/sglang/srt/openai_api/adapter.py
index 988cb2e0d..c24979ae8 100644
--- a/python/sglang/srt/openai_api/adapter.py
+++ b/python/sglang/srt/openai_api/adapter.py
@@ -1193,6 +1193,7 @@ def v1_chat_generate_request(
         top_logprobs_nums = top_logprobs_nums[0]
         modalities_list = modalities_list[0]
         lora_paths = lora_paths[0]
+        request_ids = request_ids[0]
     else:
         if tokenizer_manager.model_config.is_multimodal:
             # processor will need text input
@@ -1429,7 +1430,9 @@ async def v1_chat_completions(
         return create_error_response("Invalid request body, error: ", str(e))
     all_requests = [ChatCompletionRequest(**request_json)]
     created = int(time.time())
-    adapted_request, request = v1_chat_generate_request(all_requests, tokenizer_manager)
+    adapted_request, request = v1_chat_generate_request(
+        all_requests, tokenizer_manager, request_ids=[all_requests[0].rid]
+    )
 
     if adapted_request.stream:
         parser_dict = {}
diff --git a/python/sglang/srt/openai_api/protocol.py b/python/sglang/srt/openai_api/protocol.py
index c37442248..0cdab70cd 100644
--- a/python/sglang/srt/openai_api/protocol.py
+++ b/python/sglang/srt/openai_api/protocol.py
@@ -392,6 +392,9 @@ class ChatCompletionRequest(BaseModel):
     stream_reasoning: bool = True
     chat_template_kwargs: Optional[Dict] = None
 
+    # The request id.
+    rid: Optional[Union[List[str], str]] = None
+
     # For PD disaggregation
     bootstrap_host: Optional[str] = None
     bootstrap_port: Optional[int] = None