diff --git a/openai_api.py b/openai_api.py
index 7225562..37e1833 100644
--- a/openai_api.py
+++ b/openai_api.py
@@ -3,7 +3,7 @@
 # Usage: python openai_api.py
 # Visit http://localhost:8000/docs for documents.
 
-
+import json
 import time
 import torch
 import uvicorn
@@ -135,7 +135,7 @@ async def predict(query: str, history: List[List[str]], model_id: str):
         finish_reason=None
     )
     chunk = ChatCompletionResponse(model=model_id, choices=[choice_data], object="chat.completion.chunk")
-    yield "{}".format(chunk.json(exclude_unset=True, ensure_ascii=False))
+    yield json.dumps(chunk.model_dump(exclude_unset=True), ensure_ascii=False)
 
     current_length = 0
 
@@ -152,7 +152,7 @@ async def predict(query: str, history: List[List[str]], model_id: str):
             finish_reason=None
         )
         chunk = ChatCompletionResponse(model=model_id, choices=[choice_data], object="chat.completion.chunk")
-        yield "{}".format(chunk.json(exclude_unset=True, ensure_ascii=False))
+        yield json.dumps(chunk.model_dump(exclude_unset=True), ensure_ascii=False)
 
 
     choice_data = ChatCompletionResponseStreamChoice(
@@ -161,7 +161,7 @@ async def predict(query: str, history: List[List[str]], model_id: str):
         finish_reason="stop"
     )
     chunk = ChatCompletionResponse(model=model_id, choices=[choice_data], object="chat.completion.chunk")
-    yield "{}".format(chunk.json(exclude_unset=True, ensure_ascii=False))
+    yield json.dumps(chunk.model_dump(exclude_unset=True), ensure_ascii=False)
     yield '[DONE]'