ChatGPT Sites publishing was down for about 4 hours 38 minutes on Nov. 13
An OpenAI-side outage stopped paid users from publishing or updating ChatGPT Sites. Here is the timeline, what to do if it hit you, and how to avoid being stuck next time.
The short version
- Publishing in ChatGPT Sites failed for Plus, Pro, Business and Enterprise users on November 13, 2026.
- Timeline in UTC: started 01:17, identified 03:59, resolved 05:55.
- OpenAI's status page named no cause, and the failure produced no specific error code.
- Existing sites stayed up; only new publishes and updates were blocked.
- The advice was to publish once, wait for the all-clear, then retry a single time.
Why it matters
If a launch, portfolio or client page depends on a publish button, an outage on the provider's side can leave you stuck through no fault of your own. Knowing where the status page is, and keeping your content in a file you own, turns a stressful morning into a minor delay.
What happened on November 13
As dev.to reported, in a write-up by the troubleshooting site DebugRadar, OpenAI logged an incident on November 13, 2026 in which publishing in ChatGPT Sites failed. ChatGPT Sites is the feature that lets you put pages live from ChatGPT. The problem affected users on the Plus, Pro, Business and Enterprise plans, who could not publish new sites or push updates to existing ones.
The timeline, all in UTC, runs like this. The incident started at 01:17, OpenAI said it had identified the problem at 03:59, and it was marked resolved at 05:55. That adds up to about 4 hours 38 minutes of disruption.
OpenAI listed the incident on its status page under the title "Sites: Publishing failing", describing elevated errors on publishing. It did not say what caused them.
What it meant for your site
The write-up is clear on one point: this was not something users broke. The fault sat on OpenAI's side, so nothing on your computer or in your page's content was responsible.
There was also no data loss. While publishing was down, a site that was already live stayed frozen at its last good version, so visitors would still have seen that. Only changes you tried to push during the outage were blocked.
The failure did not come with a specific error code, either. If you were trying to publish that morning and simply saw failures, that matches the incident. The write-up also stresses that this is a separate event from earlier ChatGPT Sites problems, including an October 20 bug, a September 16 HTTP 409 error and an August 5 HTTP 500 error. If you have an older issue, it is probably not this one.
What to do when publishing fails
The write-up lays out a short routine, and it is sensible for any hosted tool.
- Check the status page first. Open status.openai.com and look for a Sites incident. If one is open, the problem is not yours to fix.
- Do not hammer the button. Every attempt goes through the same broken path. Repeated failed publishes can also leave you unsure which version is live.
- Retry once after the all-clear. Once the incident is marked resolved, publish a single time and check that the site updates.
- If it still fails, clear your browser cache and try one more time. If that fails too, treat it as a new problem rather than the old incident.
The advice boils down to patience over effort. Troubleshooting your own setup during a provider outage burns time and can muddy which version of your page is live.
Plan for the next outage
The broader lesson is about dependence. If a time-sensitive page can only go live through one button in one product, any outage on that product becomes your outage. The write-up suggests keeping a second AI workspace ready so one failure does not stop your work. It names Claude Pro and Gemini Advanced as options, though it flags that link as an affiliate link that earns the site a commission.
Our own advice from today's newsletter is simpler and cheaper: keep your content in a document or folder you control. Then a hosted tool failing costs you time, not work. If a launch, a portfolio page or a client deliverable relies on a publishing step, leave some slack in the schedule and know where the status page is before you need it.
What we still do not know
Some caution is in order. This account comes from a third-party write-up, which cites OpenAI's status page along with two outage-tracking sites, Pingoru and ModelBank. For the official record, go to OpenAI's own status page.
The biggest gap is the cause. OpenAI named none, so we cannot say whether it was a software change, a capacity problem or something else. Nor do we know how many users were affected. What the record does support is the timeline, the plans involved and the fact that the fix came from OpenAI, not from users.
What to watch
- Whether OpenAI ever explains what caused the Sites publishing failures.
- Whether OpenAI's status page shows more Sites incidents after this one.
Try it
Bookmark status.openai.com today, and keep a copy of any page you plan to publish in a document or folder on your own computer.
Sources and links
- Reportingdev.to: the article we sourced
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.
Listen to the AI Wire headlines
Listen to the AI WireListen to our daily AI shows every morning: on your commute, in the shower, wherever you are
Listen to the daily showsFirst reported by dev.to · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
