diff --git a/server_python/app.py b/server_python/app.py
index 715d70f..d4c06b6 100644
--- a/server_python/app.py
+++ b/server_python/app.py
@@ -140,6 +140,8 @@ async def chat_endpoint(request: Request):
                         temperature=temperature
                     )
 
+                    full_content = ""  # 用于累计完整内容
+
                     for idx, response in enumerate(responses):
                         if response.status_code == 200:
                             # 检查响应是否包含预期的内容
@@ -156,6 +158,29 @@ async def chat_endpoint(request: Request):
                                 'content' in response.output.choices[0]['message']):
 
                                 content = response.output.choices[0]['message']['content']
+
+                                # 只有当内容发生变化时才发送增量
+                                if len(content) > len(full_content):
+                                    delta_content = content[len(full_content):]
+                                    full_content = content
+
+                                    if delta_content.strip():  # 只有当有非空白新内容时才发送
+                                        # 构建 SSE 数据块
+                                        data = {
+                                            "id": f"chatcmpl-{uuid.uuid4()}",
+                                            "object": "chat.completion.chunk",
+                                            "created": int(datetime.utcnow().timestamp()),
+                                            "model": model,
+                                            "choices": [
+                                                {
+                                                    "index": 0,
+                                                    "delta": {"content": delta_content},
+                                                    "finish_reason": None
+                                                }
+                                            ]
+                                        }
+
+                                        yield f"data: {json.dumps(data, ensure_ascii=False)}\n\n"
                             # 否则尝试从 output.text 获取内容（DashScope特定格式）
                             elif (hasattr(response, 'output') and
                                   response.output and
@@ -163,26 +188,28 @@ async def chat_endpoint(request: Request):
 
                                 content = response.output.get('text')
 
-                            if content:
-                                # 构建 SSE 数据块
-                                data = {
-                                    "id": f"chatcmpl-{uuid.uuid4()}",
-                                    "object": "chat.completion.chunk",
-                                    "created": int(datetime.utcnow().timestamp()),
-                                    "model": model,
-                                    "choices": [
-                                        {
-                                            "index": 0,
-                                            "delta": {"content": content} if content else {},
-                                            "finish_reason": None
-                                        }
-                                    ]
-                                }
+                                # 只有当内容发生变化时才发送增量
+                                if len(content) > len(full_content):
+                                    delta_content = content[len(full_content):]
+                                    full_content = content
 
-                                yield f"data: {json.dumps(data)}\n\n"
-                            else:
-                                # 如果响应中没有内容，跳过
-                                continue
+                                    if delta_content.strip():  # 只有当有非空白新内容时才发送
+                                        # 构建 SSE 数据块
+                                        data = {
+                                            "id": f"chatcmpl-{uuid.uuid4()}",
+                                            "object": "chat.completion.chunk",
+                                            "created": int(datetime.utcnow().timestamp()),
+                                            "model": model,
+                                            "choices": [
+                                                {
+                                                    "index": 0,
+                                                    "delta": {"content": delta_content},
+                                                    "finish_reason": None
+                                                }
+                                            ]
+                                        }
+
+                                        yield f"data: {json.dumps(data, ensure_ascii=False)}\n\n"
                         else:
                             # 错误处理
                             error_data = {
@@ -193,7 +220,7 @@ async def chat_endpoint(request: Request):
                                     "code": response.code
                                 }
                             }
-                            yield f"data: {json.dumps(error_data)}\n\n"
+                            yield f"data: {json.dumps(error_data, ensure_ascii=False)}\n\n"
                             break
 
                     # 发送结束信号
@@ -210,9 +237,8 @@ async def chat_endpoint(request: Request):
                             }
                         ]
                     }
-                    yield f"data: {json.dumps(finish_data)}\n\n"
+                    yield f"data: {json.dumps(finish_data, ensure_ascii=False)}\n\n"
                     yield "data: [DONE]\n\n"
-
                 except Exception as e:
                     error_data = {
                         "error": {
@@ -220,7 +246,7 @@ async def chat_endpoint(request: Request):
                             "type": "server_error"
                         }
                     }
-                    yield f"data: {json.dumps(error_data)}\n\n"
+                    yield f"data: {json.dumps(error_data, ensure_ascii=False)}\n\n"
 
             return StreamingResponse(event_generator(), media_type="text/event-stream")
         else:
@@ -272,7 +298,7 @@ async def chat_endpoint(request: Request):
                             "totalTokens": response.usage.total_tokens
                         }
 
-                    return JSONResponse(content=chat_response)
+                    return JSONResponse(content=chat_response, ensure_ascii=False)
                 else:
                     raise HTTPException(
                         status_code=500,
diff --git a/src/services/api.ts b/src/services/api.ts
index 2751df1..7c63954 100644
--- a/src/services/api.ts
+++ b/src/services/api.ts
@@ -164,6 +164,12 @@ class ChatApi {
         if (match) {
           try {
             const data = JSON.parse(match[1]);
+            // 检查是否有完成原因，如果是完成则跳出
+            const finishReason = data.choices?.[0]?.finish_reason;
+            if (finishReason && finishReason !== "null") {
+              break;
+            }
+
             const content = data.choices?.[0]?.delta?.content;
             if (content) {
               yield content;