Replies: 44 comments 63 replies
-
|
I'd definitely also appreciate some clarification on the codex-cli limits for the Plus plan |
Beta Was this translation helpful? Give feedback.
-
|
ChatGpt Plus (GPT-5 Medium)
|
Beta Was this translation helpful? Give feedback.
-
|
Now I'm wondering what that cap would be if we use the api pricing? Would it make more sense to have Plus and fall back to API or to use Pro |
Beta Was this translation helpful? Give feedback.
-
|
Hi @shanecowherd , This is exactly what I was wondering, so I made myself two .bat files that automatically switches to the (pay as you go) API key mode when i hit the usage limit and back to the Plusplan + GPT login mode once they let me 'out of jail' and my free usage resets. Happy to show how it works if you want. |
Beta Was this translation helpful? Give feedback.
-
|
@brainit-consulting can you share some kinda gist of this? |
Beta Was this translation helpful? Give feedback.
-
📊 Token Usage
• Input: 1491682
• Output: 6819
• Total: 1498501ChatGPT Plus, one conversation, limit triggered. |
Beta Was this translation helpful? Give feedback.
-
|
Well here comes a new record 📊 Token Usage
• Input: 15927
• Output: 140
• Total: 1606716K and being limited. |
Beta Was this translation helpful? Give feedback.
-
|
On Pro, last night I coded and coded, totally solved my problem, when I was done I saw: • Input: 2249742 (+ 35030528 cached) Never hit a limit |
Beta Was this translation helpful? Give feedback.
-
|
it's way too low. it did 6 tasks and then stopped mid task 7 of 9. btw those 7 tasks were base skeleton only, no implementation. teams subscription |
Beta Was this translation helpful? Give feedback.
-
|
I'd really like to have a more clear info on how limits reset. It seems that I've hit weekly limit, and I have no idea when it will reset (Plus subscription). |
Beta Was this translation helpful? Give feedback.
-
|
i felt the same this week. My codebase was already large so i don't know what could it be |
Beta Was this translation helpful? Give feedback.
-
|
Would be nice to have a warning like how Claude Code does... "Approaching Usage Limit" or something like that. |
Beta Was this translation helpful? Give feedback.
-
|
I am totally puzzeled. I got Plus. It seems that limits in the web codex and codex cli are totally seperate. I have never run into the limit in web codex. I just now ran into the limit in codex cli, but web codex still accepts prompts . |
Beta Was this translation helpful? Give feedback.
-
|
Codex web counts as chat gpt usage not openai tokens like codex cli |
Beta Was this translation helpful? Give feedback.
-
|
+1 on what @guywilsonjr said |
Beta Was this translation helpful? Give feedback.
-
|
Codex Cloud should not be free anymore per doc https://developers.openai.com/codex/pricing 
|
Beta Was this translation helpful? Give feedback.
-
|
aw dang it :'( so its not free anymore |
Beta Was this translation helpful? Give feedback.
-
|
Limits are better but still too low. |
Beta Was this translation helpful? Give feedback.
-
|
Here we go again. |
Beta Was this translation helpful? Give feedback.
-
|
I just hit the usage limits with Plus and I get this message.
Usually it was try again in 5 hours... Is there a new usage limit? |
Beta Was this translation helpful? Give feedback.
-
|
Well, you may hit the weekly limitation. |
Beta Was this translation helpful? Give feedback.
-
|
I hit the weekly about every 2 days, about 12h per day usage. Have to use a friend's account to continue (not rich). But even combined that's not enough to cover 7x8h days. |
Beta Was this translation helpful? Give feedback.
-
Literally the same, never hit any 5 hours limit, then hit the weekly limit straightly.... |
Beta Was this translation helpful? Give feedback.
-
|
Exactly! The funny thing is that in previous weeks I used it much more. I was even hitting the 5-hour limits. But this week I only used it once on Monday, and then suddenly I hit the weekly limit. I was also thinking maybe there are monthly limits too, but then again it’s strange that it says it refreshes in 3 days. The limits and usage should definitely be explained in more detail. Unless they aren’t fixed and just get adjusted depending on how much you use it. |
Beta Was this translation helpful? Give feedback.
-
|
What they're doing is borderline criminal. |
Beta Was this translation helpful? Give feedback.
-
|
Compared to Claude, OpenAI already offers quite a lot for the money. But of course, some people don't need the 3,000 thinking requests in ChatGPT. So why can't they use it via Codex? It's legitimate to question this. In the end, a middle ground will have to be found. |
Beta Was this translation helpful? Give feedback.
-
|
Last week I hit the 5h limit once, and the weekly limit on Sunday evening (had to wait 11 hours). Tokens do cost money and we can do things on our side to reduce usage ("cognitive load" for the LLMs):
Of course it depends on the project, but I'm trying to chunk my tasks in a way to have faster responses which usually means less tokens usage. |
Beta Was this translation helpful? Give feedback.
-
If you have to work to accomodate the AI to save tokens, why are you using AI at all then? It's there so help you reach your goal/vision, not the other way around. And Currently 73€ is not $20 (re:linked post). 1 teams + 2 plus should last about 5 1/2 days. But I work every day. API cost is ridiculous. $1250 for 1M tokens. I'm not rich, but I'm working on it, and that's why I have to spend to gain velocity. Of course it can't be free, but this is predatory. It's on them to reduce their costs and also not destory our environment while expanding and improving and operating. I have now skipped git commits, even if the summary is so nice. Edit: It's $1.25 (why are they writing it like $1.2500?) for 1M tokens |
Beta Was this translation helpful? Give feedback.
-
|
Yeah even following my own rules, and being extremely careful, I hit the Teams weekly limit Wednesday evening... It would be nice to see how much usage we have left per day/per month. |
Beta Was this translation helpful? Give feedback.
-
|
I'm on the Plus plan and just started using Codex VSCode, first session. Worked fine for half a day, then I got the 5h limit and it asked me to wait for 30 mins. Fine with me, no problem, I have some other things to do anyway. So far so good. But one day later I hit another limit (this is only the second time I've hit any kind of limit), this time it asks me to wait for 5 days! 5 days, you gotta be kidding! Not a few hours, but 5 days! A tool quitting all of a sudden, out of the blue, for that long is crossing a red line for me. Totally unacceptable. I wouldn't mind waiting for a few hours, but waiting for days renders this tool totally unworkable in any practical setting. 5h limits are ok, but the weekly limits are nuts. Please fix the limits ASAP! |
Beta Was this translation helpful? Give feedback.
-
|
Just use the API. The 5-hour limit corresponds to approximately €2-3 in API costs. |
Beta Was this translation helpful? Give feedback.
-
|
Exactly: when I ran out of limits, I used the API and consumed about €2 per day. |
Beta Was this translation helpful? Give feedback.
-
How do you rotate? I purchased a new Plus subscription. Logged out everywhere and logged in with the new account. Still has the chat history in VS Code and also the usage error. I even uninstalled Codex and reinstalled it in VS Code. Still the same. There seems to be some issue with the session of the old account not being cleared. |
Beta Was this translation helpful? Give feedback.
-
|
I'm logged in with my main account. create scripts in /usr/local/sbin/ /usr/local/sbin/codex-main #!/bin/env bash
cp ~/.codex/auth.json.main ~/.codex/auth.json/usr/local/sbin/codex-second #!/bin/env bash
cp ~/.codex/auth.json.second ~/.codex/auth.json/usr/local/sbin/codex-third #!/bin/env bash
cp ~/.codex/auth.json.third ~/.codex/auth.jsonBefore starting codex I run e.g. |
Beta Was this translation helpful? Give feedback.
-
|
Ah, thanks for the reply. I think you are using CLI, not the VS Code Codex Extension. It seems the credentials are stored differently there. I do not even have the .codex folder... :-/ Maybe I need to switch to CLI. |
Beta Was this translation helpful? Give feedback.
-
|
Let me stick with CC |
Beta Was this translation helpful? Give feedback.
-
|
I opened a feature request addressing some of the discussed issues in a different way: The proposal is to let Codex CLI continue a session after hitting plan limits by allowing re-authentication — entering a new API key or logging in with another account — instead of terminating. This would prevent losing context or leaving code in an inconsistent state. Vote for it if you like it. |
Beta Was this translation helpful? Give feedback.
-
|
I mean can't you just pay and set up an account with 5 dollars to allow finishing the last prompt? |
Beta Was this translation helpful? Give feedback.
-
|
the VS Code extension now also shows usage: |
Beta Was this translation helpful? Give feedback.
-
Does anyone know if quota consumption varies depending on the model used? example :
|
Beta Was this translation helpful? Give feedback.
-
|
Yes each model has different pricing. |
Beta Was this translation helpful? Give feedback.
-
|
I have a similar issue how do I get in contact with support for subscriptions? |
Beta Was this translation helpful? Give feedback.
-
|
I think OpenAi is in for a cash grab , you need to have 2 separate payment tiers , one for GPT and one for codex , on plus plan you hit quota in 3 hours and than there is that message " you are fuqked up , come back in 5 days 17 hours " , what is worse now after 6 and half days codex is still non functional ,even i have api with cash tied , just some dumb loader in codex tab , no options ,no message, no nothing . Open AI is just scam scheme obviously |
Beta Was this translation helpful? Give feedback.
-
|
Yeah dark pattern they could fix but aren't. And always the buy out 230€ super expensive Pro sub. |
Beta Was this translation helpful? Give feedback.
-
|
I agree, im thinking to use claude but have to see what it can do . Tried kilo code (pure trash.. ) , tried Roo Code ( row for row copy paste Kilo code, trash) ... went out of options to have some ai for checking my code |
Beta Was this translation helpful? Give feedback.
-
|
fwiw qwen is free, at least I have not met any limits, not great but not terrible either. |
Beta Was this translation helpful? Give feedback.
-
|
There is no point to 5 hours limits. That exist on Claude Code because it used to have almost no limits, thus 5 hour limit (session) and 50 session/mo limit was necessary. Now everything has been rate limited, no point to 5 hour limits (in CODEX, Claude, anywhere) if weekly limits are very very low. Just let people use their limits in one session (which is what is actually happening now). |
Beta Was this translation helpful? Give feedback.
-
|
Trying the factory ai droid free trial plan, it's so liberating not having to "watch over your shoulder" expecting being limited any moment now. A subscription should allocate however many (current guess around 20-25M per month *10 for Pro) tokens per month and there should not be any 5h limits. |
Beta Was this translation helpful? Give feedback.
-
|
you can use API keys and pay for exactly what you use. You can even use a different provider if you please. |
Beta Was this translation helpful? Give feedback.
-
|
We're aware. This is not what this is about |
Beta Was this translation helpful? Give feedback.
-
|
Does the weekly limit reset at a fixed time or is it dynamic based on usage? |
Beta Was this translation helpful? Give feedback.
-
Beta Was this translation helpful? Give feedback.
-
|
This is not a place to advertise your stuff. Especially not with bot generated content |
Beta Was this translation helpful? Give feedback.
-
|
I've gotten some mileage out of running longer prompts through a few passes of GPT5 asking it to phrase for efficiency and clarity as a Codex input. |
Beta Was this translation helpful? Give feedback.
-
|
hit my first ever limits in codex cloud last night, sad times |
Beta Was this translation helpful? Give feedback.
-
|
yup it's complete bullshit |
Beta Was this translation helpful? Give feedback.
-
|
back to copy paste to gpt5 until that runs out.. |
Beta Was this translation helpful? Give feedback.
-
|
I have a Plus plan, hobby coding usage, and already hit the 5h limit twice in a couple days. I used to use it much more without hitting a limit. Is this a temporary issue? |
Beta Was this translation helpful? Give feedback.
-
|
they suck you in then raise the price |
Beta Was this translation helpful? Give feedback.
-
|
I stopped using it since complaining last week. Turns out it doesn’t take much to run coding LLM locally with qwen models, and it’s faster than I was getting with codex. It isn’t as polished of an experience but I think that’s partly a new user issue I’ll sort out over time. |
Beta Was this translation helpful? Give feedback.
-
|
good one, open ai should only break even, was orig a non-profit |
Beta Was this translation helpful? Give feedback.
-
|
@james633 Can you tell more about your local setup with Qwen? |
Beta Was this translation helpful? Give feedback.
-
LM Studio, add a model, it will tell you which ones are recommended for your setup, then there’s a built in github tool and you just paste a github token into a setup file and it ingests it and BAM github integration. GPT walked me through most of it. There’s a bit of calibration you have to do through trial and error on how many tokens you can manage at a time, but for the most part it ends up being one new conversation per request. |
Beta Was this translation helpful? Give feedback.
-
|
It’s really frustrating. Last night around 7:00 PM, it showed that the 5-hour limit was 100% used, and said it would reset at 00:21. I logged into Codex via the browser dashboard to check my usage, and there it shows 5 hours remaining = 100% and weekly remaining = 66%. It’s really strange — the same usage data (5 h and weekly) is interpreted completely differently: |
Beta Was this translation helpful? Give feedback.
-
|
Exactly same response here. I thought my API key had been leaked and it was being used while I slept. The 5-hour reset time changes, but the usage does not. It remains at 100% and has been since yesterday. Completely unusable. Regardless of, I revoked all access to codex and logged out every session, but it seems like this is by design and there was a strong limit reduction. |
Beta Was this translation helpful? Give feedback.
-
|
Agree it’s not just frustrating it’s unusable. Zero visibility, zero understanding of when or why, and having no way to know whether it will spend 10 minutes considering and reconsidering a small change means you can never know how much usage to expect even if you knew how much you had left. |
Beta Was this translation helpful? Give feedback.
-
Hi bro, I found the reason. This is a product bug — the server was sending the remaining usage to the plugin, but the plugin mistakenly interpreted it as used amount. |
Beta Was this translation helpful? Give feedback.
-
|
I realized that a few seconds after making my comment, however, I stopped using Codex because most requests would simply fail and halt the entire chain of thought mid request, even with 100% pure availability. They literally flipped the quota usage. Went back to Roocode as I have much more control over the quality of the output. In addition, I had to delete my cache, reinstall plugin, reauthenticate, and migrate my chat folder to get it responding again. Only using it as PR review for now. |
Beta Was this translation helpful? Give feedback.
-
|
same here. really weird, especially since i'm on the $200 plan. this has never happened before, but it says i'm already over 90% usage, even though i haven't changed my coding habits. plus, my $200 anthropic account is already limited by the week too. i'm so disappointed with these policies and how these companies are handling things right now. |
Beta Was this translation helpful? Give feedback.
-
|
If we weren’t paying for this then the irony of all the jokes about AI overlords might be amusing in that here we are in our first interactions with AI, and in the perfect example of a totalitarianism, we have no way of knowing if or when we have violated the rules, but are punished for our violations nonetheless. |
Beta Was this translation helpful? Give feedback.
-
|
well they train us to be dep on ai, so we just have to go back to gpt and hand coding or fork out the cash |
Beta Was this translation helpful? Give feedback.
-
|
maybe they all going into price fixing |
Beta Was this translation helpful? Give feedback.
-
|
Hi bro, I found the reason. This is a product bug — the server was sending the remaining usage to the plugin, but the plugin mistakenly interpreted it as used amount. |
Beta Was this translation helpful? Give feedback.
-
|
new model 
https://community.openai.com/t/codex-updates-mini-model-higher-limits-priority-processing/1365540 |
Beta Was this translation helpful? Give feedback.
and others
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Is the limits for the Plus tier in the ChatGPT app the same as the limits in codex? (3000 Thinking/week)
Beta Was this translation helpful? Give feedback.
All reactions