In response to feedback from subscribed users regarding the swift depletion of their OpenAI Codex usage quotas, Thibault Sottiaux, the leader of the OpenAI Codex & ChatGPT team, shared a statement on the X platform. He confirmed that three key issues impacting usage calculations have been pinpointed: Firstly, inefficient image handling and context compression during extended sessions; Secondly, unexpectedly high usage levels in the top 5% (P95 and above) of computer history records; Thirdly, the conversation title generation feature consuming more resources than initially estimated. OpenAI has promptly formed a specialized team to delve into these problems and intends to deploy fixes by approximately 14:00 Pacific Time the following day, at which point it will also reset the usage quotas for all subscribed users. Furthermore, the company is set to embark on developing a comprehensive new plan to enhance usage efficiency starting next week.