OpenAI logged three separate incidents on 30 September, touching ChatGPT Pages, Pro and Plus access, and API latency. The longest of the three, affecting ChatGPT Space Pages, ran for 45 hours 27 minutes, a fault on OpenAI's own side rather than anything you could have fixed downstream.
Started
02:20 UTC
Duration
Lasted 1d 21h
Source
IsDown
Next time
OpenAI is degraded. Is your product working?
An incident on their side does not always mean an outage on yours, and the only way to know is to already be watching your own endpoints. Set that up in under a minute and you will be notified the next time your own site is affected. And there will be a next time.
50 monitors for your own websites. No card.
Affected components: APIs, Chat, Responses, ChatGPT, Web
What happened?
The day opened at 02:20 UTC with errors in ChatGPT Space Pages, where OpenAI said users could see failures creating or interacting with Pages, including Page tools and live Page sessions. That one ran the longest of the three, at 45 hours 27 minutes, with OpenAI applying a mitigation by 06:03, reporting full recovery by 10:34, then having to apply mitigations again at 15:00 and warning at 16:51 that some users might still see slow or timed out results creating new pages or spaces. A second, shorter incident started at 05:47 UTC, affecting ChatGPT Pro and Plus users with elevated errors across impacted services. OpenAI applied a mitigation within nine minutes and called it recovered by 10:36, a total span of 7 hours 38 minutes. A third incident began later, at 17:05 UTC, with elevated latency on some API requests, specifically flagging the Responses and Chat Completions APIs. OpenAI applied an initial mitigation at 17:51, noted improvement, and moved to monitoring by 18:21, closing out at 3 hours 32 minutes. Five components were listed as affected across the day: ChatGPT, Web, APIs, Chat and Responses. No user reports were filed during this window, so everything on this page comes from OpenAI's own status updates, eleven in total across the three incidents. No post-mortem has been published for any of the three, so none of the updates explain what caused the errors or the latency, only what was done about them.
Learning
OpenAI's incidents this day sat inside its own infrastructure, the Pages feature, account tier routing for Pro and Plus, and API request handling, none of which an external check against a simple endpoint would reliably catch. A probe that just confirms OpenAI's API is reachable can return green while specific features like Space Pages or specific request types like Responses and Chat Completions are degraded for real users, because the fault lives in a narrower slice of the service than a basic reachability check exercises. That is the shape to recognise in your own product: a top-level health check passing does not mean every feature path behind it is working, especially when a vendor is applying mitigations in stages, as OpenAI did twice within the Pages incident alone. With 97 incidents logged in the past 90 days at OpenAI and a median duration of 1 hour 52 minutes, elevated error periods are a frequent fact of running on this provider, not a rare event. Your own monitoring covers your half of the stack, and UptimeRobot now watches the provider's half too, so next time something breaks you know which side it was without guessing.
