The English region is about to get OpenAI's Codex blasted, because lately quite a lot of people using the ChatGPT Codex tool suddenly found that their quota was being consumed unusually fast.
For example, previously they could use up their quota within a week, but now it might be gone in just one or two days—even in the official client. So over the past few days, everyone has been complaining that OpenAI must have secretly cut their quota.
Currently, Tibo, who at OpenAI is responsible for Codex, posted a response saying: we haven’t quietly changed the quota. Many people who say their quota is acting strangely are actually using a tool called sub2api (a shared interface).
Sure enough, after this guy posted the tweet, the comment section still wasn’t buying it—because many people haven’t used sub2api at all. They’ve only been using the official app properly, and their quota has still been shrinking.
And the community widely suspects that maybe the model itself is consuming more resources, or there’s a bug in the backend statistics—anyway, users’ quotas are getting reduced.
For example, previously they could use up their quota within a week, but now it might be gone in just one or two days—even in the official client. So over the past few days, everyone has been complaining that OpenAI must have secretly cut their quota.
Currently, Tibo, who at OpenAI is responsible for Codex, posted a response saying: we haven’t quietly changed the quota. Many people who say their quota is acting strangely are actually using a tool called sub2api (a shared interface).
Sure enough, after this guy posted the tweet, the comment section still wasn’t buying it—because many people haven’t used sub2api at all. They’ve only been using the official app properly, and their quota has still been shrinking.
And the community widely suspects that maybe the model itself is consuming more resources, or there’s a bug in the backend statistics—anyway, users’ quotas are getting reduced.