I am a pro member. In the latest 0.50.x version, it seems that after the 500 times quota is used up, you can only get a response through the auto mode. If you use any specified model, it will always be in a timeout and unresponsive state. The auto mode only supports gpt4.1, which is not very helpful for real projects and cannot work like claude3.7 or gemini 2.5 pro. If this continues in the future, I will cancel the renewal of my pro membership.