diff --git a/lightllm/server/api_tgi.py b/lightllm/server/api_tgi.py
index 3e1b3733c..f9b070c3c 100755
--- a/lightllm/server/api_tgi.py
+++ b/lightllm/server/api_tgi.py
@@ -1,3 +1,4 @@
+import os
 import collections
 from typing import AsyncGenerator
 from fastapi import BackgroundTasks, Request
@@ -8,6 +9,8 @@
 from .httpserver.manager import HttpServerManager
 import json
 
+RETURN_LIST = os.getenv("RETURN_LIST", "FALSE").upper() in ["ON", "TRUE", "1"]
+
 
 def format_tgi_params(params, num_beam: int = 1):
     """
@@ -132,7 +135,10 @@ async def tgi_generate_impl(request: Request, httpserver_manager: HttpServerMana
     if return_details:
         ret["details"]["beam_sequences"] = beam_sequences
     # wrap generation inside a Vec to match api-inference
-    json_compatible_item_data = jsonable_encoder([ret])
+    if RETURN_LIST:
+        json_compatible_item_data = jsonable_encoder([ret])
+    else:
+        json_compatible_item_data = jsonable_encoder(ret)
     return JSONResponse(content=json_compatible_item_data)