update pay ui

mcp_server.py

@@ -65,31 +65,37 @@ MODEL_CONFIGS = {
         "api_key": "sk-7363bdb28d7d4bf0aa68eb9449f8f063",
         "base_url": "https://api.deepseek.com",
         "model": "deepseek-chat",  # 默认模型
+        "max_tokens": 8192,  # DeepSeek 限制为 8192
     },
     "kimi-k2": {
         "api_key": "sk-TzB4VYJfCoXGcGrGMiewukVRzjuDsbVCkaZXi2LvkS8s60E5",
         "base_url": "https://api.moonshot.cn/v1",
         "model": "moonshot-v1-8k",  # 快速模型
+        "max_tokens": 8192,  # moonshot-v1-8k 限制为 8k
     },
     "kimi-k2-thinking": {
         "api_key": "sk-TzB4VYJfCoXGcGrGMiewukVRzjuDsbVCkaZXi2LvkS8s60E5",
         "base_url": "https://api.moonshot.cn/v1",
         "model": "kimi-k2-thinking",  # 深度思考模型
+        "max_tokens": 32768,  # Kimi 思考模型支持更大
     },
     "glm-4.6": {
         "api_key": "",  # 需要配置智谱AI密钥
         "base_url": "https://open.bigmodel.cn/api/paas/v4",
         "model": "glm-4",
+        "max_tokens": 8192,
     },
     "deepmoney": {
         "api_key": "",  # 空值
         "base_url": "http://111.62.35.50:8000/v1",
         "model": "deepmoney",
+        "max_tokens": 32768,
     },
     "gemini-3": {
         "api_key": "",  # 需要配置Google API密钥
         "base_url": "https://generativelanguage.googleapis.com/v1",
         "model": "gemini-pro",
+        "max_tokens": 8192,
     },
 }
 
@@ -2451,11 +2457,13 @@ class MCPAgentIntegrated:
 
             try:
                 # 尝试使用选中的模型流式 API
+                # 从模型配置获取 max_tokens，默认 8192
+                model_max_tokens = model_config.get("max_tokens", 8192) if model_config else 32768
                 stream = planning_client.chat.completions.create(
                     model=planning_model,
                     messages=messages,
                     temperature=1.0,
-                    max_tokens=32768,
+                    max_tokens=model_max_tokens,
                     stream=True,  # 启用流式输出
                 )
 
@@ -3673,6 +3681,8 @@ async def stream_role_response(
 
         # 第一次调用：可能触发工具调用
         tool_calls_made = []
+        # 从模型配置获取 max_tokens，默认 8192
+        max_tokens = model_config.get("max_tokens", 8192)
         if openai_tools:
             response = client.chat.completions.create(
                 model=model_config["model"],
@@ -3681,7 +3691,7 @@ async def stream_role_response(
                 tool_choice="auto",
                 stream=False,  # 工具调用不使用流式
                 temperature=0.7,
-                max_tokens=32768,  # 增大 token 限制以避免输出被截断
+                max_tokens=max_tokens,
             )
 
             assistant_message = response.choices[0].message
@@ -3741,7 +3751,7 @@ async def stream_role_response(
             messages=messages,
             stream=True,
             temperature=0.7,
-            max_tokens=8192,  # 大幅增加 token 限制以避免输出被截断
+            max_tokens=max_tokens,
         )
 
         full_content = ""