On 29 September 2026, OpenAI had two separate incidents starting within three minutes of each other: support tools went down, and ChatGPT, Codex and the API including the Agents API returned elevated errors. The longer of the two ran 5h 31m, and both were OpenAI's own infrastructure failing, not anything on your end.
Started
17:49 UTC
Duration
Lasted 5h 31m
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, APIs, File uploads, Voice mode, GPTs, Image Generation, Deep Research, Operator, Codex, Chat, Compliance, Images, Embeddings, Batch, Fine-tuning, Files, Audio, Realtime, Responses, Moderations, Login, Web, Search
What happened?
OpenAI logged two incidents on 29 September 2026, starting at 17:49 and 17:52 UTC respectively, close enough together that they likely shared a root cause even though OpenAI tracked them as separate items. The first affected the Help Center and support chat: OpenAI's own update at 17:49 said both were unavailable and told users to email support@openai.com instead. Six minutes later OpenAI narrowed that to say the Help Center had come back but support chat was still down. That incident ran for 5h 31m before OpenAI applied a mitigation and moved to monitoring at 21:50. The second incident, filed three minutes after the first, described elevated errors across ChatGPT, Codex and the API including the Agents API. OpenAI's own text said some users would see failed requests, trouble logging in or signing up, and tasks that would not complete. That description stayed essentially unchanged through three updates at 17:52, 18:24 and 19:08, with OpenAI only adding the Agents API to the list of affected surfaces along the way. Mitigation and monitoring followed at 19:39, putting that incident's duration at 5h 23m. The affected components list for the day spans 12 of OpenAI's 23 tracked components, including ChatGPT, APIs, File uploads, Voice mode, GPTs, Image Generation, Deep Research, Operator, Codex, Chat, Compliance and Images, though not every update named every one of these explicitly. No cause is given for either incident beyond "mitigation applied". OpenAI has not published a post-mortem for either item, and no user reports were filed during the window.
Learning
OpenAI's own status page is the only record here, and it describes failures at the application and API layer, not at the network edge where a simple uptime ping would catch anything. Requests failing, logins not completing and tasks hanging are symptoms that live inside OpenAI's request handling, somewhere a basic check against your own endpoints would never see, because your endpoints were working fine, the problem was in what OpenAI's servers did once a request reached them. Over the last 90 days OpenAI has logged 96 such incidents with a median duration of 1 hour 52 minutes, and across its history it averages about 15 incidents a month, so this kind of elevated-error episode is a recurring shape rather than a one-off. If you only monitor your own integration points, a day like this can pass with your dashboards green while your users still can't get a response out of ChatGPT or the API. Your monitoring covers your half, UptimeRobot now watches the provider's half too, so next time you know which side broke without guessing.
