diff --git a/apps/application/flow/step_node/ai_chat_step_node/impl/base_chat_node.py b/apps/application/flow/step_node/ai_chat_step_node/impl/base_chat_node.py
index 67999f37314..27cbe9e9cf7 100644
--- a/apps/application/flow/step_node/ai_chat_step_node/impl/base_chat_node.py
+++ b/apps/application/flow/step_node/ai_chat_step_node/impl/base_chat_node.py
@@ -523,9 +523,10 @@ def generate_prompt_question(self, prompt, model):
             images = self._process_images(image)
         if video and vision:
             videos = self._process_videos(video, model)
-        return HumanMessage(
-            content=[*videos, *images, {"type": "text", "text": self.workflow_manage.generate_prompt(prompt)}]
-        )
+        prompt = self.workflow_manage.generate_prompt(prompt)
+        if images or videos:
+            return HumanMessage(content=[*videos, *images, {"type": "text", "text": prompt}])
+        return HumanMessage(content=prompt)
 
     def is_vision(self):
         if "vision" in self.node_params_serializer.data: