diff --git a/backend/app/backend.py b/backend/app/backend.py
index bbf33e3..e1b0d52 100644
--- a/backend/app/backend.py
+++ b/backend/app/backend.py
@@ -151,27 +151,62 @@ async def chat_endpoint(
         raise HTTPException(status_code=400, detail="message required")
 
     thread_id = request.thread_id or str(uuid.uuid4())
-    result = await agent_service.process_message(
-        request.message, thread_id, request.model, request.user_id
-    )
     
-    # 提取 token 统计信息
-    token_usage = result.get("token_usage", {})
-    input_tokens = token_usage.get('prompt_tokens', token_usage.get('input_tokens', 0))
-    output_tokens = token_usage.get('completion_tokens', token_usage.get('output_tokens', 0))
-    elapsed_time = result.get("elapsed_time", 0.0)
-    
-    actual_model = request.model if request.model in agent_service.graphs else next(iter(agent_service.graphs.keys()))
-    
-    return ChatResponse(
-        reply=result["reply"],
-        thread_id=thread_id,
-        model_used=actual_model,
-        input_tokens=input_tokens,
-        output_tokens=output_tokens,
-        total_tokens=input_tokens + output_tokens,
-        elapsed_time=elapsed_time
-    )
+    try:
+        result = await agent_service.process_message(
+            request.message, thread_id, request.model, request.user_id
+        )
+        
+        # 提取 token 统计信息
+        token_usage = result.get("token_usage", {})
+        input_tokens = token_usage.get('prompt_tokens', token_usage.get('input_tokens', 0))
+        output_tokens = token_usage.get('completion_tokens', token_usage.get('output_tokens', 0))
+        elapsed_time = result.get("elapsed_time", 0.0)
+        
+        actual_model = request.model if request.model in agent_service.graphs else next(iter(agent_service.graphs.keys()))
+        
+        return ChatResponse(
+            reply=result["reply"],
+            thread_id=thread_id,
+            model_used=actual_model,
+            input_tokens=input_tokens,
+            output_tokens=output_tokens,
+            total_tokens=input_tokens + output_tokens,
+            elapsed_time=elapsed_time
+        )
+    except Exception as e:
+        error(f"同步响应异常: {e}")
+        
+        # === 兜底输出机制 ===
+        error_message = str(e)
+        is_timeout_error = any(keyword in error_message.lower() for keyword in 
+                               ["timeout", "timed out", "超时", "connection", "unavailable", "不可用"])
+        
+        # 1. 自我介绍
+        intro_text = "你好！我是 AI 智能助手，我可以帮你处理各种问题，包括查询通讯录、词典翻译、新闻分析、知识库检索、联网搜索等。\n\n"
+        
+        # 2. 错误信息（红色突出）
+        error_display = f"**⚠️ 当前遇到问题**\n\n```diff\n- {error_message}\n```\n\n"
+        
+        # 3. 模型切换提示（如果是超时/不可用错误）
+        switch_hint = ""
+        if is_timeout_error:
+            switch_hint = "💡 **提示**：当前模型可能响应超时或不可用，请尝试手动切换到其他模型（如 DeepSeek、智谱AI等）。\n\n"
+        
+        # 4. 组合完整兜底回复
+        fallback_text = intro_text + error_display + switch_hint
+        
+        actual_model = request.model if request.model in agent_service.graphs else next(iter(agent_service.graphs.keys()))
+        
+        return ChatResponse(
+            reply=fallback_text,
+            thread_id=thread_id,
+            model_used=actual_model,
+            input_tokens=0,
+            output_tokens=0,
+            total_tokens=0,
+            elapsed_time=0.0
+        )
 
 # ========== 历史查询接口 ==========
 @app.get("/threads")
@@ -225,7 +260,34 @@ async def chat_stream_endpoint(
             yield "data: [DONE]\n\n"
         except Exception as e:
             error(f"流式响应异常: {e}")
-            yield f"data: {json.dumps({'type': 'error', 'message': str(e)}, ensure_ascii=False)}\n\n"
+            
+            # === 兜底输出机制 ===
+            error_message = str(e)
+            is_timeout_error = any(keyword in error_message.lower() for keyword in 
+                                   ["timeout", "timed out", "超时", "connection", "unavailable", "不可用"])
+            
+            # 1. 自我介绍
+            intro_text = "你好！我是 AI 智能助手，我可以帮你处理各种问题，包括查询通讯录、词典翻译、新闻分析、知识库检索、联网搜索等。\n\n"
+            
+            # 2. 错误信息（红色突出）
+            error_display = f"**⚠️ 当前遇到问题**\n\n```diff\n- {error_message}\n```\n\n"
+            
+            # 3. 模型切换提示（如果是超时/不可用错误）
+            switch_hint = ""
+            if is_timeout_error:
+                switch_hint = "💡 **提示**：当前模型可能响应超时或不可用，请尝试手动切换到其他模型（如 DeepSeek、智谱AI等）。\n\n"
+            
+            # 4. 组合完整兜底回复
+            fallback_text = intro_text + error_display + switch_hint
+            
+            # 5. 以 llm_token 方式发送兜底回复，模拟打字机效果
+            for char in fallback_text:
+                yield f"data: {json.dumps({'type': 'llm_token', 'node': 'fallback', 'token': char}, ensure_ascii=False)}\n\n"
+                import asyncio
+                await asyncio.sleep(0.01)
+            
+            # 6. 发送错误事件
+            yield f"data: {json.dumps({'type': 'error', 'message': error_message}, ensure_ascii=False)}\n\n"
             yield "data: [DONE]\n\n"
     
     return StreamingResponse(