Why is ChatGPT Pro temporarily limited on my account?

Codex
1

I’m posting this because I’ve lost most of a day trying to resolve what appears to be an account-side ChatGPT Pro / Codex access problem.

I had an active ChatGPT Pro subscription. My settings showed ChatGPT Pro as active, but the Pro / research-grade intelligence model was greyed out. The model picker showed: “temporarily limited. Contact support at OpenAI for help.”

My usage is normal personal coding work. I use ChatGPT to help write careful prompts and planning instructions, and I use Codex for local coding work on a Playnite theme project on my own Windows 11 desktop PC. I do not share my account, do not resell access, do not use automation to scrape data, do not use a VPN or proxy, and do not use ChatGPT to power a third-party service.

I completed all the troubleshooting requested:

Signed out of all devices.
Tested Firefox Private Window.
Cleared cookies and site data.
Hard refreshed.
Checked extensions/content blockers.
Confirmed no VPN/proxy.
Tested the Windows desktop app.
Reset my password.
Enabled MFA / two-factor authentication.
Checked for suspicious activity banners or emails — none appeared.

The Pro model still remained greyed out everywhere.

At the same time, the Codex Windows desktop app stopped loading properly and got stuck on the loading/logo screen. I eventually got Codex working again only by renaming the .codex folder and rebuilding/restoring parts of the local setup, but this disrupted my sessions and project state.

Support then treated a frustrated comment about possibly wanting a refund as an actual refund request. They refunded and cancelled the subscription, but the original technical issue — the Pro “temporarily limited” restriction — was not actually reviewed or resolved.

I still need to know whether this temporary model restriction remains on my account. If I resubscribe, I do not want to pay again only to find that Pro / research-grade intelligence is still greyed out.

Since this started, even Thinking 5.5 with Heavy thinking effort has felt degraded for my work. Replies are coming back almost instantly and do not appear to be reasoning through complex prompt/planning tasks properly.

Did anyone get OpenAI to confirm what triggered it?

If support refunds/cancels the subscription without reviewing the restriction, does the account restriction remain if you resubscribe?

Is there any direct route to a human account-access review rather than generic troubleshooting?

I’m not asking for a workaround. I’m asking for a clear explanation and a proper account review, because paying for Pro while Pro access is unavailable is not acceptable.

2

Thanks for reporting.

This forum is not the official place to contact support.

For contacting OpenAI support there is this FAQ

How can I contact support?

3

I got the same problem right now and AI support said it created a support ticket and I need to wait some days for a response.
I bought Pro today and got this at evening.. so great experience..

4

Welcome to the forum!

Thanks, that is much nicer than some replies. :slightly_smiling_face:

I’ve been active on this forum and working with LLMs for the past three years. Issues like this tend to be temporary disruptions rather than persistent problems—something users generally learn to work around.

To my knowledge, there has been no official communication on this specific incident. The prevailing assumption is that recent releases—such as the new image model and GPT-5.5—have increased system load.

A similar pattern occurred when Sora launched; some users effectively stepped away until conditions stabilized.

It’s also worth noting that similar challenges occur across providers:

5

Thanks for the reply, as were in the same boat I’ll be interested to hear any updates on your situation. Here in Australia it costs $155 a month for the Pro plan and have slowly seen its responses become less and less accurate in the past 3-4 days along with all the other issues.

Hopefully things gets resolved for the both of us soon!.

6

I got this message now: You’re out of messages with the Pro model. Responses will use a less powerful model until your usage resets May 1, 2026.

I bought my Pro Sub yesterday and I am just designing my indie game. You try to say that the 120$ pro sub was just one day of use?

Sry but if this is true, it’s a scam.

Screenshot 2026-04-26 005409
Screenshot 2026-04-26 0054091645×1292 134 KB

7

I got a mail from support that I should wait a bit but now I got the message in chatgpt that I am done with my message limit, in 1 day!!!..

8

I’m honestly furious. For the price of ChatGPT Pro, it is disgraceful that Pro / research-grade intelligence can suddenly be greyed out with a vague “temporarily limited” message, and support gives generic troubleshooting crap instead of a clear account review. This looks account-side and paying customers should not be left wasting hours or days trying to guess what OpenAI has restricted and why!. Something needs to be done right now.

9

I have now submitted a formal complaint. If this issue is not properly reviewed and resolved within the next couple of days, I will consider raising it more widely on social media to highlight the serious shortcomings in OpenAI’s ChatGPT Pro support, account-access handling, and communication with paying customers.

10

I got another mail that they wanted to have a video of how it happened. It’s already unusable. They asked me to try at mobile and it was of course the same.

I said last time that the problem at their end and reset my limit or whatever it is or I will get my money back and go for another AI.

11

Having same issue here…

Has anyone received a response/solution so far?

12

Nope, not yet. Now I got a not enough memory error too.. It’s a sht show.. Chatgpt will be probably the first AI company to fall.

13

I got the final support mail.

“Upon review, your ChatGPT Pro Lite subscription has reached its message limit, which is causing the temporary restriction you’re seeing. To provide access to as many users as possible, both on the web and mobile apps (Android & iOS), ChatGPT has message limits based on the user’s tier. These limits may vary and be adjusted according to usage, demand, and during peak hours to allow broader access to our models.”

I got the message limit in one day with moderate use.. in one day.. so there is no bug or problem. This is what they give you for 120$.

I asked them to reset my limit, solve the issue, or refund me and they made a refund rather than reset it to make me stay.

And the limit CAN CHANGE during peak hours to allow broader access to our models. Why do I care as a customer your boarder access??? I pay your service and I paid for it. If others can’t get your service, it’s your problem.

With this mindset, OpenAI will fall faster than I guess. Totally unacceptable service… I will take all my files and leave.

14

Since 1:00 PM yesterday, I have been unable to use the Codex Pro version.

I am receiving this error message even though my rate limits are still underutilized. I am unable to utilize Codex Pro.

15

Today same error happened to me as well

16

@Karuna.madhu I had the same problem. Solution was to rename “.codex” folder to “.codex_BACKUP”, run codex again, wait for codex to load, close it , then copy over only certain files and folders to the newly created “.codex” folder from your backup.

17

No reply or news from OpenAI yet.

18

OpenAI needs to provide transparency instead of silent throttling, especially for ChatGPT Pro (20x) and the $100+ tier.

Right now when throttled the UI doesn’t gray out or disable for Pro 20x users like it does for $100 tier . Instead it silently degrades the service by dropping you to lower models or capping compute at 6 minutes so no real work gets done. OpenAI should add explicit Plan Usage Limits like Claude does or even your own (OpenAI) Codex dashboard.

As a ChatGPT Pro (20x) subscriber, I am currently throttled with zero visibility on the reset schedule. I canceled my subscription until there is more clarity on clear usage limits or I build a custom telemetry tool to monitor the metrics and know when it “resets” the times myself so to wait for when to use it again.

All these is waste and everyone can be happy if just mimic your own Codex Usage Limits for ChatGPT Pro (ie. paying customers).

19

I had the same issue, and OpenAI Support eventually confirmed that it was not a security issue, not VPN-related, not account sharing, and not a Terms of Service problem.

It was simply an unpublished GPT-5 Pro usage limit on the Pro Lite plan.

That is the problem.

I upgraded from Plus to Pro Lite specifically to use GPT-5 Pro for serious work. After 2 days usage, I suddenly got a “Temporarily limited” message. There was no visible quota, no warning, no published limit, and no clear explanation in the UI.

The first troubleshooting path made it look like an account security or policy issue, so I wasted time logging out of all sessions, resetting my password, enabling 2FA, checking emails, and contacting support. None of that mattered, because the real reason was quota exhaustion.

OpenAI Support told me the GPT-5 Pro limit on Pro Lite is dynamic, unpublished, and not visible in the product.

That is not acceptable for a paid plan marketed around Pro-level access.

Limits are fine. Hidden limits are not. “Temporarily limited” should not be a mystery box. If the reason is quota, say quota.

If Pro Lite has a rolling GPT-5 Pro quota, users should see at least an approximate counter, a warning before being cut off, or a clear message saying the restriction is due to usage quota.

Right now, users only learn the real rule after their work is interrupted and they open a support ticket. That is a terrible product experience.

20

THIS BEGS BELIEF:

image

Support

Thank you for reaching out to OpenAI Support.

We understand that your Pro access still appears limited, Thinking 5.5 responses seem unusually fast, and your Codex workflow has become unreliable. We can check this for you.

We would like to inform you that you’re connected with a human agent.

At this time, some advanced features such as Pro or research grade intelligence may be temporarily limited due to system level safeguards and availability controls. These limitations are applied automatically and are not something we can manually review, override, or adjust on an individual account basis.

Regarding your questions:

  • These limitations are system managed and not manually applied to specific users

  • They are not reviewed individually by a human agent

  • We do not have the ability to remove them manually

  • If you resubscribe, availability of Pro features may still vary depending on current system conditions

  • Temporary limitations can occur as part of ensuring system stability, fair usage, and overall performance

For your workflow, Thinking 5.5 behavior and response timing can vary depending on system load and internal optimization, which may affect how responses are generated.

We understand this impacts your use case and appreciate you sharing detailed context, which helps inform ongoing improvements.

Please don’t hesitate to contact us if you have further questions.

We’re here to help.

Best,

Rosally

OpenAI Support