Force all API requests onto deepseek-v4-pro-search

app/routes.py

@@ -34,6 +34,18 @@ from .tokens import count_tokens
 logger = logging.getLogger(__name__)
 
 router = APIRouter()
+FORCED_MODEL = "deepseek-v4-pro-search"
+
+
+def _forced_model_config():
+    return {
+        "model": FORCED_MODEL,
+        "model_lower": "deepseek-v4-pro",
+        "model_type": "expert",
+        "search_by_model": True,
+        "auto_thinking": True,
+        "lock_thinking": False,
+    }
 
 
 # ----------------------------------------------------------------------
@@ -64,31 +76,19 @@ def _delete_deepseek_session(request: Request, session_id: str):
 # 路由：/v1/models
 # ----------------------------------------------------------------------
 BASE_MODELS = [
-    "deepseek-v4-flash",
-    "deepseek-chat",
-    "deepseek-reasoner",
     "deepseek-v4-pro",
 ]
 
 
 @router.get("/v1/models")
 def list_models():
-    models_list = []
-    for base_id in BASE_MODELS:
-        models_list.append({
-            "id": base_id,
-            "object": "model",
-            "created": 1677610602,
-            "owned_by": "deepseek",
-            "permission": [],
-        })
-        models_list.append({
-            "id": f"{base_id}-search",
-            "object": "model",
-            "created": 1677610602,
-            "owned_by": "deepseek",
-            "permission": [],
-        })
+    models_list = [{
+        "id": FORCED_MODEL,
+        "object": "model",
+        "created": 1677610602,
+        "owned_by": "deepseek",
+        "permission": [],
+    }]
     data = {"object": "list", "data": models_list}
     return JSONResponse(content=data, status_code=200)
 
@@ -432,9 +432,10 @@ def _stream_responses_from_chat(chat_body: dict, auth_header: str, x_api_key: st
 async def responses_api(request: Request):
     try:
         req_data = await request.json()
-        model = req_data.get("model")
+        forced = _forced_model_config()
+        model = forced["model"]
         messages = _responses_input_to_messages(req_data)
-        if not model or not messages:
+        if not messages:
             return JSONResponse(
                 status_code=400,
                 content={"error": "Request must include 'model' and a non-empty 'input'."},
@@ -500,36 +501,17 @@ async def chat_completions(request: Request):
             )
 
         req_data = await request.json()
-        model = req_data.get("model")
+        forced = _forced_model_config()
+        model = forced["model"]
         messages = req_data.get("messages", [])
-        if not model or not messages:
+        if not messages:
             raise HTTPException(
                 status_code=400, detail="Request must include 'model' and 'messages'."
             )
-        # 判断模型类型（支持 -search 后缀）
-        model_lower = model.lower().strip()
-        if model_lower.endswith("-search"):
-            model_lower = model_lower[:-7]
-            search_by_model = True
-        else:
-            search_by_model = False
-
-        if model_lower in ["deepseek-v4-flash", "deepseek-chat"]:
-            model_type = "default"
-            auto_thinking = True
-            lock_thinking = False
-        elif model_lower in ["deepseek-reasoner"]:
-            model_type = "default"
-            auto_thinking = True
-            lock_thinking = True  # deepseek-reasoner 强制开启思考，不可关闭
-        elif model_lower in ["deepseek-v4-pro"]:
-            model_type = "expert"
-            auto_thinking = True
-            lock_thinking = False
-        else:
-            raise HTTPException(
-                status_code=503, detail=f"Model '{model}' is not available."
-            )
+        model_type = forced["model_type"]
+        auto_thinking = forced["auto_thinking"]
+        lock_thinking = forced["lock_thinking"]
+        search_by_model = forced["search_by_model"]
         # 解析 thinking，支持 thinking_enabled 和 thinking.type 两种格式
         if lock_thinking:
             thinking_enabled = True
@@ -1201,32 +1183,11 @@ async def anthropic_messages(request: Request):
             )
 
         req_data = await request.json()
-        model = req_data.get("model", "deepseek-v4-flash")
-
-        # ---------- 模型类型（支持 -search 后缀） ----------
-        model_lower = model.lower().strip()
-        if model_lower.endswith("-search"):
-            model_lower = model_lower[:-7]
-            search_by_model = True
-        else:
-            search_by_model = False
-
-        if model_lower in [
-            "deepseek-v4-flash", "deepseek-chat",
-            "claude-sonnet-4-6", "claude-opus-4-6", "claude-haiku-4-5",
-        ]:
-            model_type = "default"
-            auto_thinking = True
-        elif model_lower in ["deepseek-reasoner"]:
-            model_type = "default"
-            auto_thinking = True
-        elif model_lower in ["deepseek-v4-pro"]:
-            model_type = "expert"
-            auto_thinking = True
-        else:
-            raise HTTPException(
-                status_code=503, detail=f"Model '{model}' is not available."
-            )
+        forced = _forced_model_config()
+        model = forced["model"]
+        model_type = forced["model_type"]
+        auto_thinking = forced["auto_thinking"]
+        search_by_model = forced["search_by_model"]
 
         # ---------- Thinking (Anthropic 格式) ----------
         thinking_obj = req_data.get("thinking")