OpenAI logged three separate incidents on 9 October 2026, touching the Compliance API, the ChatGPT Android app, and workspace admin plugin controls. The longest ran 70 hours 40 minutes, a delay in Costs data through the Compliance API that OpenAI's own updates repeatedly said did not affect ChatGPT usage or API Platform requests. All three were OpenAI side faults.
Started
17:23 UTC
Duration
Lasted 2d 22h
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: ChatGPT, Web, APIs, Compliance
What happened?
The day's biggest incident started at 17:23 UTC and ran 70 hours 40 minutes, OpenAI's longest-running item on the Compliance API that day. It began as delayed Costs data publishing through the Compliance API, and OpenAI's own text changed slightly between the first two updates, first naming the "Costs API" then correcting itself to the "Compliance API" four minutes later. Throughout, OpenAI stated that ChatGPT usage and API Platform requests were not affected and that other reporting channels remained available. By 18:46 UTC OpenAI said it had identified the cause, though it never named it publicly. A mitigation went in at 22:03 UTC, with OpenAI warning that the backlog catch-up would continue through Sunday 11 October and that some customers might still see incomplete Costs data in the meantime. Separately, at 21:00 UTC, some ChatGPT Android users began seeing a "This content is unavailable in this version of the app" message in place of certain content. OpenAI said a fix was rolling out gradually in the latest Android build and pointed affected users to the web app as a workaround. That incident lasted 3 hours 26 minutes. A third incident started at 21:48 UTC, affecting workspace administrators' ability to view and manage plugin availability and settings, with some users seeing plugins as unavailable entirely. OpenAI reported recovery by 22:52 UTC, 1 hour 38 minutes after it opened. No user reports were filed during the window for any of the three incidents. Four components in total show affected status across the day: APIs, Compliance, ChatGPT and Web.
Learning
None of these three incidents were the kind of thing an external uptime check against your own ChatGPT or API integration would catch, because OpenAI itself said core usage paths stayed up throughout. The Compliance API delay sat in a reporting and billing data pipeline, not in the request path your product depends on to function, so your own monitoring could stay green while OpenAI's Costs data was stale for days. The Android content error and the plugin management issue were both narrower still, tied to a specific client version or a specific admin workflow rather than the service as a whole. This is the general shape of a provider-side partial fault: the top-level service answers fine, but a secondary feature, a reporting surface, or an admin control breaks underneath it, and nothing you can probe from outside will tell you that's happening. Watching your own endpoints only tells you your half is up. OpenAI's history here, 109 incidents in 90 days with a median duration of 1 hour 50 minutes, shows this kind of partial fault is routine for them, not a one-off. Your monitoring covers your half, UptimeRobot now watches the provider's half too, so next time you know which side broke without guessing.
