Merge pull request #3 from omegaduncan/main

fix: Improve round-robin provider selection and add Claude max_tokens auto-fill

main.py

@@ -182,11 +182,13 @@ async def request_model(self, request: RequestModel, token: str):
 
         # 检查是否启用轮询
         api_index = api_list.index(token)
-        use_round_robin = False
-        auto_retry = False
+        use_round_robin = True
+        auto_retry = True
         if config['api_keys'][api_index].get("preferences"):
-            use_round_robin = config['api_keys'][api_index]["preferences"].get("USE_ROUND_ROBIN")
-            auto_retry = config['api_keys'][api_index]["preferences"].get("AUTO_RETRY")
+            if config['api_keys'][api_index]["preferences"].get("USE_ROUND_ROBIN") == False:
+                use_round_robin = False
+            if config['api_keys'][api_index]["preferences"].get("AUTO_RETRY") == False:
+                auto_retry = False
 
         return await self.try_all_providers(request, matching_providers, use_round_robin, auto_retry)
 
@@ -207,6 +209,7 @@ async def try_all_providers(self, request: RequestModel, providers: List[Dict],
                 else:
                     raise HTTPException(status_code=500, detail="Error: Current provider response failed!")
 
+
         raise HTTPException(status_code=500, detail=f"All providers failed: {request.model}")
 
 model_handler = ModelRequestHandler()

request.py

@@ -450,6 +450,13 @@ async def get_vertex_claude_payload(request, engine, provider):
         "system": system_prompt or "You are Claude, a large language model trained by Anthropic.",
     }
 
+    # 檢查是否需要添加 max_tokens
+    if 'max_tokens' not in payload:
+        if "claude-3-5-sonnet" in model:
+            payload['max_tokens'] = 8192
+        elif "claude-3" in model:  # 處理其他 Claude 3 模型
+            payload['max_tokens'] = 4096
+
     miss_fields = [
         'model',
         'messages',