diff --git a/mcp_server.py b/mcp_server.py
index 2e47acec..9c3eef89 100644
--- a/mcp_server.py
+++ b/mcp_server.py
@@ -1999,7 +1999,7 @@ class MCPAgentIntegrated:
             model=self.kimi_model,
             messages=messages,
             temperature=1.0,  # Kimi 推荐
-            max_tokens=8192,  # 足够容纳 reasoning_content
+            max_tokens=128000,  # 足够容纳 reasoning_content
         )
 
         choice = response.choices[0]
@@ -2074,7 +2074,7 @@ class MCPAgentIntegrated:
                 model=self.deepmoney_model,
                 messages=messages,
                 temperature=0.7,
-                max_tokens=8192,
+                max_tokens=128000,
             )
 
             summary = response.choices[0].message.content
@@ -2268,7 +2268,7 @@ class MCPAgentIntegrated:
                 model="kimi-k2-turbo-preview",  # 使用非思考模型，更快
                 messages=messages,
                 temperature=0.7,
-                max_tokens=8192,  # 增加 token 限制以支持图表配置
+                max_tokens=128000,  # 增加 token 限制以支持图表配置
             )
 
             summary = response.choices[0].message.content
@@ -2355,7 +2355,7 @@ class MCPAgentIntegrated:
                 model=self.deepmoney_model,
                 messages=messages,
                 temperature=0.3,
-                max_tokens=4096,
+                max_tokens=32768,
             )
 
             title = response.choices[0].message.content.strip()
@@ -2450,7 +2450,7 @@ class MCPAgentIntegrated:
                     model=planning_model,
                     messages=messages,
                     temperature=1.0,
-                    max_tokens=8192,
+                    max_tokens=32768,
                     stream=True,  # 启用流式输出
                 )
 
@@ -2494,7 +2494,7 @@ class MCPAgentIntegrated:
                             model=self.deepmoney_model,
                             messages=messages,
                             temperature=0.7,
-                            max_tokens=8192,
+                            max_tokens=32768,
                         )
 
                         plan_content = fallback_response.choices[0].message.content
@@ -2690,7 +2690,7 @@ class MCPAgentIntegrated:
                         model="kimi-k2-turbo-preview",
                         messages=messages,
                         temperature=0.7,
-                        max_tokens=8192,
+                        max_tokens=32768,
                         stream=True,  # 启用流式输出
                     )
 
@@ -2724,7 +2724,7 @@ class MCPAgentIntegrated:
                                 model=self.deepmoney_model,
                                 messages=messages,
                                 temperature=0.7,
-                                max_tokens=8192,
+                                max_tokens=32768,
                             )
 
                             final_summary = fallback_response.choices[0].message.content
@@ -3676,7 +3676,7 @@ async def stream_role_response(
                 tool_choice="auto",
                 stream=False,  # 工具调用不使用流式
                 temperature=0.7,
-                max_tokens=8192,  # 增大 token 限制以避免输出被截断
+                max_tokens=32768,  # 增大 token 限制以避免输出被截断
             )
 
             assistant_message = response.choices[0].message