diff --git a/ruoyi-fastapi-backend/module_admin/controller/ragflow_controller.py b/ruoyi-fastapi-backend/module_admin/controller/ragflow_controller.py
index 36db12c..10eb13f 100644
--- a/ruoyi-fastapi-backend/module_admin/controller/ragflow_controller.py
+++ b/ruoyi-fastapi-backend/module_admin/controller/ragflow_controller.py
@@ -8,7 +8,7 @@ import json
 import time
 from typing import Any, Dict, Generator, Optional, Tuple
 
-from fastapi import APIRouter, Depends, Request
+from fastapi import APIRouter, BackgroundTasks, Depends, Request
 from fastapi.responses import StreamingResponse
 
 from module_admin.service.login_service import LoginService
@@ -96,6 +96,7 @@ def remove_style_hint(question: str) -> Tuple[str, bool]:
 @ragflowController.post('/converse_with_chat_assistant')
 async def converse_with_chat_assistant(
     request: Request,
+    background_tasks: BackgroundTasks,
     converse_params: ConverseWithChatAssistantModel,
 ):
     """
@@ -202,20 +203,21 @@ async def converse_with_chat_assistant(
         store_hash = get_question_hash(cache_question)
         logger.info(f'[RAG_CACHE] 准备存储 | chat_id={converse_params.chat_id} | question={cache_question} | hash={store_hash}')
         
-        async def make_cache_store(chat_id: str, question: str):
-            async def store_answer(answer: str):
-                if redis and answer and len(answer.strip()) >= 10:
-                    try:
-                        await _async_store_qa(chat_id, question, answer, redis)
-                        logger.info(f'[RAG_CACHE] 语义缓存存储成功 | chat_id={chat_id} | answer_length={len(answer)}')
-                    except Exception as e:
-                        logger.warning(f'语义缓存存储失败: {e}')
-            return store_answer
+        answer_queue = asyncio.Queue()
         
-        cache_store = await make_cache_store(converse_params.chat_id, cache_question)
+        async def cache_store_background():
+            try:
+                answer = await answer_queue.get()
+                if redis and answer and len(answer.strip()) >= 10:
+                    await _async_store_qa(converse_params.chat_id, cache_question, answer, redis)
+                    logger.info(f'[RAG_CACHE] 语义缓存存储成功 | chat_id={converse_params.chat_id} | answer_length={len(answer)}')
+            except Exception as e:
+                logger.warning(f'[RAG_CACHE] 语义缓存存储失败: {e}')
+        
+        background_tasks.add_task(cache_store_background)
         
         return StreamingResponse(
-            stream_ragflow_response(result, converse_params.chat_id, start_time, cache_store_func=cache_store),
+            stream_ragflow_response(result, converse_params.chat_id, start_time, answer_queue=answer_queue),
             media_type='text/event-stream',
             headers={
                 'Cache-Control': 'no-cache',
@@ -230,7 +232,7 @@ async def converse_with_chat_assistant(
         return ResponseUtil.error(msg=f'对话服务异常: {str(e)}')
 
 
-def stream_ragflow_response(result: Generator, chat_id: str, start_time: float, cache_store_func=None) -> Generator[str, None, None]:
+def stream_ragflow_response(result: Generator, chat_id: str, start_time: float, answer_queue=None) -> Generator[str, None, None]:
     """
     流式处理RAGFlow响应 - 同步版本，修复首token延迟
     """
@@ -323,13 +325,17 @@ def stream_ragflow_response(result: Generator, chat_id: str, start_time: float,
         total_time = time.perf_counter() - start_time
         logger.info(f'[RAG_SERVER {time.time():.3f}] ⏱️ Total Stream Duration ({chat_id}): {total_time:.3f}s')
         
-        if cache_store_func and last_answer and len(last_answer.strip()) >= 10:
+        if answer_queue and last_answer and len(last_answer.strip()) >= 10:
             try:
                 import asyncio
-                asyncio.run(cache_store_func(last_answer))
-                logger.info(f'[RAG_CACHE] 缓存存储完成 | chat_id={chat_id} | answer_length={len(last_answer)}')
+                loop = asyncio.get_event_loop()
+                if loop.is_running():
+                    asyncio.run_coroutine_threadsafe(answer_queue.put(last_answer), loop)
+                else:
+                    loop.run_until_complete(answer_queue.put(last_answer))
+                logger.info(f'[RAG_CACHE] 已将答案放入队列 | chat_id={chat_id} | answer_length={len(last_answer)}')
             except Exception as cache_err:
-                logger.warning(f'[RAG_CACHE] 缓存存储失败: {cache_err}')
+                logger.warning(f'[RAG_CACHE] 队列写入失败: {cache_err}')
 
 
 # 聊天助手列表