Usage cap for GPT-4 and message error bugs

Closing this duplicate topic.

Please use: