It is just me, or has Codex started gouging or just plain broken.
Yesterday I paid $48 to purchase extra credits for a critical project this weekend.
However this morning is has reported that just today it has not only used all my 5 hour budget from my Plus subscription, but also all of the credits I purchased - and I’ve only put 8 queries to it (x4 agents).
Typically I have been able to submit around 100 (x4) queries per day. The 8 queries were no different in scope or general description than any other I have put for the past 6 months and it’s also working on exactly the same codebase.
Obviously with this I will most definitely be cancelling my subscription to OpenAI and will move to a different coder as I can’t rely on them when the chips are down, it’s pretty irritating as this weeked is the culmination of years of work for which I will be set back months if I can’t pass this hurdle, but is there a way to at least get a refund on this crappy money-burn by Codex?
Check this post. Others are having issues too. Open AI support commented and said they had a problem with output tokens calculation and are rolling back some changes to resolve. They will reset peoples usage for who was affected.
You need to check the posting before from me. It happened Friday already, then it seems, OpenAI rolled it back on Saturday. Now today on Sunday, it is like on Friday again with the 3-4 requests before you hit the limit
Been saying this since their mess up on Thursday and they claimed it was a “bug”, hilarious they couldn’t smoke test something like this.
I found an alternative and it’s been great. If OpenAI can’t fix this, my sub will expire this month – sad because I’ve been subbed for almost a year and have been using Codex since launch. I really don’t think OpenAI cares either, it’s seems to be about the money for them on this one.
Same on Plus. Since yesterday’s Codex “Usage” rollout, limits feel broken: 8 tasks already count as 60% of my weekly cap, and 3 cloud tasks drained the entire 5-hour window. I’ve hammered Codex for six months and never hit limits — even with heavier workloads. This looks like a metering regression/misclassification, not intended behavior. So, I hope this is not new reality.
What the hell, just finished my “weekly usage” !
Lol… in 2 days + i bougth a 5$ more token and can’t use it… what’s the point ! (And i’m suppose to have 100k tokens …)
