diff --git a/ruoyi-fastapi-backend/module_admin/controller/ragflow_controller.py b/ruoyi-fastapi-backend/module_admin/controller/ragflow_controller.py
index 0fbaf87..9b68d83 100644
--- a/ruoyi-fastapi-backend/module_admin/controller/ragflow_controller.py
+++ b/ruoyi-fastapi-backend/module_admin/controller/ragflow_controller.py
@@ -92,7 +92,13 @@ async def converse_with_chat_assistant(
         if converse_params.stream:
             return StreamingResponse(
                 stream_ragflow_response(result, converse_params.chat_id, start_time),
-                media_type='text/event-stream'
+                media_type='text/event-stream',
+                headers={
+                    'Cache-Control': 'no-cache',
+                    'Connection': 'keep-alive',
+                    'X-Accel-Buffering': 'no',
+                    'Transfer-Encoding': 'chunked'
+                }
             )
         
         # 非流式响应
@@ -115,12 +121,15 @@ async def converse_with_chat_assistant(
 
 def stream_ragflow_response(result: Generator, chat_id: str, start_time: float) -> Generator[str, None, None]:
     """
-    流式处理RAGFlow响应 - 同步版本
+    流式处理RAGFlow响应 - 同步版本，修复首token延迟
     """
     last_answer = ""
     first_token_received = False
     start_stream_time = time.perf_counter()
     
+    # 立即发送连接建立消息，解决首token延迟
+    yield "event: ping\ndata: {\"status\": \"connected\"}\n\n"
+    
     try:
         for chunk in result:
             # 检查第一个token的延迟
diff --git a/ruoyi-fastapi-backend/utils/ragflow_util.py b/ruoyi-fastapi-backend/utils/ragflow_util.py
index 8908c4f..61fb26f 100644
--- a/ruoyi-fastapi-backend/utils/ragflow_util.py
+++ b/ruoyi-fastapi-backend/utils/ragflow_util.py
@@ -53,12 +53,20 @@ class RAGFlowClient:
         return result
     
     def _stream_request(self, method: str, endpoint: str, **kwargs) -> Generator[Dict[str, Any], None, None]:
-        """发送流式HTTP请求"""
+        """发送流式HTTP请求，修复SSE缓冲问题"""
         url = f"{self.base_url}{endpoint}"
         headers = kwargs.pop('headers', self.headers.copy())
         
+        # 添加防止缓冲的响应头
+        headers.update({
+            'Cache-Control': 'no-cache',
+            'Connection': 'keep-alive',
+            'X-Accel-Buffering': 'no'  # Nginx禁用缓冲
+        })
+        
         response = requests.request(method, url, headers=headers, stream=True, **kwargs)
         
+        # 立即处理响应，yield每个chunk
         for line in response.iter_lines():
             if line:
                 line = line.decode('utf-8')