Why ppl posting their BS ideas here while the question of topic is CLEAR? Why flooding? @OpenAI_Support any updates whether it’s BUG or new POLICY?
2 Likes
I believe you are a Plus/Pro user, if so you are not affected “YET”.
Business users are the ones which have a completely unusable subscription.
They mention within a few weeks, the same will be rolled out to all other users, this means that the people in this thread who are currently not complaining and think thats everythings alright, will be more then complaining.
2 Likes
@Sasha123 from my understanding its a new policy, its not a bug, its working as intended. However, as soon as these changes gets rolled out to all other users, there will be no users left as its that bad, no one will be paying for the service.
4 Likes
That doesn’t change what I’ve said. I have never even reached 70% of my limit despite coding every day (despite a small period OpenAI claimed a bug was draining rate limits)
It was inevitable that these rate limits would be tightened to reflect the actual cost of service.
It may be wise to try the API version and see how much it actually costs, and ensure that whatever you’re doing merits spending the money. Otherwise, it’s a good time to reduce dependencies to AI-for-coding.
1 Like
This isn’t even funny anymore — I couldn’t stay silent and had to respond. You haven’t experienced the issue yourself, yet you still feel entitled to comment in a way that comes across as outright gloating over those who actually have.
Just because everything works fine for you doesn’t mean other users can’t run into problems. So a fair question is — what are you even doing in this thread?
Try typing even a simple “Hello” in Codex. If that alone eats up 5% of a five-hour limit, then we can talk. Otherwise, you clearly don’t know what you’re talking about and don’t understand how serious this issue is for the people discussing it here.
3 Likes
You recently cut our weekly quota for codex included with ChatGPT in half. OK. It is what it is.
But, why on earth did you reduce the 5h to make it practically useless? I’ve always been good about not running into the 5h. Now, I can run into it very quickly. Some of us like to use codex once a week or two, and not spread it out all week.
Obviously, you are trying to pressure us to use more usage based pricing. But, all you’ve done is accelerate our need to replace codex. It’s just such a violation of trust and insensitivity to your customer base, that becoming free from OpenAI becomes a top priority.
3 Likes
idk bro inference is infinite in open ai frfr
1 Like
Yeah, it turns out are rates were reset because the codex reached a milestone of 3 million users, and will reset every million users until 10 million users, if I remember the press release I read at 4am today.
The reset only happened on Plus/Pro and Go accounts, Business accounts were NOT reset.
1 Like
Yeah, it really bothers some folks that OpenAI has been more human-centric than normal corporations, huh?
Sorry, I only know my lane, and I did mention ‘if I remember the 4 am post I read.’
I’ll try harder next time.
All good, was just trying to state the fact that it wasn’t “all” accounts which was reset, which again leads to unfairness etc etc.
3 Likes
The main point is that people affected by it shouldn’t confuse it with something that was related to the issues in this thread that have been raised.
We humans, tend to quickly assoicate such patterns too quickly.
1 Like
What I found pre-nurf was a 5 hour limit was about 4 hours of actual coding. This is without using any subagents etc. That extra hour allowed for coffee breaks etc. Now its more like 20-40 minutes if your very lucky. I think currently the shortest ive ran it out the 5 hour limit was in 11 minutes with these new limits, again no additional agents/subagents etc, just my hands and Vscode.
Tell me about it, thats why I also had to say something. I also apreshate you chiming in and its good to know some of us have support from others in the same circumstance.
1 Like
It’s been several days and it seems “human-centric“ company doesn’t give a dang about Business accounts. @OpenAI_Support we just really need to know a simple thing - is it new policy or a bug with totally crippled 5hr limits? If it’s a bug - no problem, I think we all ready to wait several more days… If it’s a new policy just let us know, most of us are ready to pay more, but for other products.
P.S. All good samaritans with their nonsense, please, don’t flood here too much, it’ll be very hard for responsible ppl to read it all because of your bragging…
P.P.S. None of my 6 accounts on business had limits resetted btw.
1 Like
IM starting to see some improvements in 5H limits now. Maybe OpenAI actually did something. Last few hours i pushed 4 deep tasks: still have 79% of 5H limits. Maybe we gonna see normalization again.
Since 3.April it was 3 deep tasks - 0%
How are others? Feeling the same?
I can’t properly test the first Business account yet, as it’s still blocked for another day and a half. On my personal Plus account, performance does appear to be improved, and by a fair margin. However, I can’t rule out some placebo effect there, given that most of the expected issues seemed to be concentrated on Business accounts.
Are you on business or plus?
@d_ul I have both business accounts and also plus account.
However my applogies, didnt see you were replying to JJ.
I have currently used numerous rounds, and have used less than 4%. Perhaps you should be trying to understand why a “hello” is costing 5% of your usage?
Again, these companies are subsidizing your token usage. This time period has been wonderful to experiment and try things out, but if you have leaned too much on letting AI do the work for you, then you can either find a way to profit off of these tokens, or learn yourself and reduce the costs.
Otherwise, you clearly don’t know what you’re talking about and don’t understand how serious this issue is for the people discussing it here.
Not sure that connects. I have used all SOTA LLM providers, along with these frameworks. I also am fully aware of how much tokens these services actually cost. We have been given a great deal, and I have been happy to take advantage of it.
Codex easily has the best rate limits currently.
Just because everything works fine for you doesn’t mean other users can’t run into problems. So a fair question is — what are you even doing in this thread?
Because this is an open forum. I use Codex daily - despite you saying otherwise - and have been happy with it. If Codex is getting more expensive, then maybe it’s worth using these plans to learn how to code, and use it for pair programming. It’s a much more affordable option