However, my actual situation is that ever since the reset yesterday, my usage consumption has been about three times higher than before. From start to finish, I have been using the proper ChatGPT login method, exclusively using Codex and Pi, with zero sub2api behavior, and yet the consumption is still incorrect.