Today, the English-speaking community is blowing up about OpenAI's Codex. The reason? Recently, a lot of people using the ChatGPT Codex tool suddenly noticed their usage quota is being consumed way faster than before. For example, what used to last a full week now runs out in just a day or two—even the official client is experiencing this. So over the past few days, everyone’s been complaining, wondering if OpenAI secretly cut the quotas. Currently, Tibo, who’s in charge of Codex at OpenAI, posted a response saying, 'We didn’t secretly change the quotas. Many of the people reporting issues with their quotas are actually using a tool called sub2api (shared API interface).' Sure enough, after he posted this tweet, the replies were still full of skepticism. Many people said they’ve never used sub2api and were just using the official app properly, yet their quotas still shrank. The community is widely speculating that it might be because the model itself is consuming more resources, or there’s a bug in the backend stats. Either way, users’ quotas are definitely getting reduced.