When the VLM API is added to dify, it doesn't work as well as it used to.

[weweliang]

Self Checks

• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
• [FOR CHINESE USERS] 请务必使用英文提交 Issue，否则会被关闭。谢谢！:）
• Please do not modify this template :) and fill in all the required fields.

Content

Siliconflow's Qwen2-VL-72B-Instruct visual model API gives different results when you use it with dify instead of calling it directly. The direct calls work better than those made through dify.