feat: llm统计现已记录模型反应时间

src/llm_models/utils.py

@@ -155,7 +155,7 @@ class LLMUsageRecorder:
             logger.error(f"创建 LLMUsage 表失败: {str(e)}")
 
     def record_usage_to_database(
-        self, model_info: ModelInfo, model_usage: UsageRecord, user_id: str, request_type: str, endpoint: str
+        self, model_info: ModelInfo, model_usage: UsageRecord, user_id: str, request_type: str, endpoint: str, time_cost: float = 0.0
     ):
         input_cost = (model_usage.prompt_tokens / 1000000) * model_info.price_in
         output_cost = (model_usage.completion_tokens / 1000000) * model_info.price_out
@@ -164,6 +164,8 @@ class LLMUsageRecorder:
             # 使用 Peewee 模型创建记录
             LLMUsage.create(
                 model_name=model_info.model_identifier,
+                model_assign_name=model_info.name,
+                model_api_provider=model_info.api_provider,
                 user_id=user_id,
                 request_type=request_type,
                 endpoint=endpoint,
@@ -171,6 +173,7 @@ class LLMUsageRecorder:
                 completion_tokens=model_usage.completion_tokens or 0,
                 total_tokens=model_usage.total_tokens or 0,
                 cost=total_cost or 0.0,
+                time_cost = round(time_cost or 0.0, 3),
                 status="success",
                 timestamp=datetime.now(),  # Peewee 会处理 DateTimeField
             )

src/llm_models/utils_model.py

@@ -71,6 +71,7 @@ class LLMRequest:
             (Tuple[str, str, str, Optional[List[ToolCall]]]): 响应内容、推理内容、模型名称、工具调用列表
         """
         # 模型选择
+        start_time = time.time()
         model_info, api_provider, client = self._select_model()
 
         # 请求体构建
@@ -105,6 +106,7 @@ class LLMRequest:
                 user_id="system",
                 request_type=self.request_type,
                 endpoint="/chat/completions",
+                time_cost=time.time() - start_time,
             )
         return content, (reasoning_content, model_info.name, tool_calls)
 
@@ -149,8 +151,6 @@ class LLMRequest:
         # 请求体构建
         start_time = time.time()
         
-        
-        
         message_builder = MessageBuilder()
         message_builder.add_text_content(prompt)
         messages = [message_builder.build()]
@@ -190,6 +190,7 @@ class LLMRequest:
                 user_id="system",
                 request_type=self.request_type,
                 endpoint="/chat/completions",
+                time_cost=time.time() - start_time,
             )
         
         if not content:
@@ -208,6 +209,7 @@ class LLMRequest:
             (Tuple[List[float], str]): (嵌入向量，使用的模型名称)
         """
         # 无需构建消息体，直接使用输入文本
+        start_time = time.time()
         model_info, api_provider, client = self._select_model()
 
         # 请求并处理返回值
@@ -228,6 +230,7 @@ class LLMRequest:
                 user_id="system",
                 request_type=self.request_type,
                 endpoint="/embeddings",
+                time_cost=time.time() - start_time,
             )
 
         if not embedding: