diff --git a/apps/common/config/embedding_config.py b/apps/common/config/embedding_config.py
index b33dd83a11d..5938797c398 100644
--- a/apps/common/config/embedding_config.py
+++ b/apps/common/config/embedding_config.py
@@ -18,7 +18,7 @@ class ModelManage:
     @staticmethod
     def get_model(_id, get_model):
         model_instance = ModelManage.cache.get(_id)
-        if model_instance is None:
+        if model_instance is None or not model_instance.is_cache_model():
             model_instance = get_model(_id)
             ModelManage.cache.set(_id, model_instance, timeout=60 * 30)
             return model_instance
diff --git a/apps/setting/models_provider/base_model_provider.py b/apps/setting/models_provider/base_model_provider.py
index 8a9ab5e2b66..022cf92e3fc 100644
--- a/apps/setting/models_provider/base_model_provider.py
+++ b/apps/setting/models_provider/base_model_provider.py
@@ -89,6 +89,10 @@ class MaxKBBaseModel(ABC):
     def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs):
         pass
 
+    @staticmethod
+    def is_cache_model():
+        return True
+
 
 class BaseModelCredential(ABC):
 
diff --git a/apps/setting/models_provider/impl/xf_model_provider/model/llm.py b/apps/setting/models_provider/impl/xf_model_provider/model/llm.py
index a9e41dbc452..a589412867e 100644
--- a/apps/setting/models_provider/impl/xf_model_provider/model/llm.py
+++ b/apps/setting/models_provider/impl/xf_model_provider/model/llm.py
@@ -20,6 +20,9 @@
 
 
 class XFChatSparkLLM(MaxKBBaseModel, ChatSparkLLM):
+    @staticmethod
+    def is_cache_model():
+        return False
 
     @staticmethod
     def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs):