OpenAI restores ChatGPT after outage

ChatGPT had a pretty practical outage on Sunday night — not the flashy kind where the whole product disappears, but the kind that breaks real work. OpenAI said users were seeing elevated errors when uploading files in ChatGPT and when creating tasks in Codex Cloud. By 10:35 PM on May 10, 2026, OpenAI marked the incident resolved and said all affected services had fully recovered. ### What actually broke? Two things, specifically. ChatGPT file uploads were failing for some users, and Codex Cloud task creation was also throwing errors. That means the problem hit a part of ChatGPT people use for documents, spreadsheets, screenshots, and other file-based prompts — plus a part of OpenAI’s coding workflow stack that developers use to kick off cloud tasks. (status.openai.com) ### When did it start and end? OpenAI’s incident page shows the issue began at 6:23 PM on Sunday, May 10. The company posted a monitoring update at 8:23 PM saying it had applied a mitigation, then marked the incident fully recovered at 10:35 PM. So the disruption lasted a little over 4 hours from first detection to resolution. (status.openai.com) ### Was all of ChatGPT down? Not exactly — and that distinction matters. OpenAI classified the event as “degraded performance,” not a full outage. The status page tied the incident to one ChatGPT component and one Codex component, which suggests a narrower failure centered on file operations and task creation rather than a complete platform collapse. The catch is that a “partial” failure can still feel total if your workflow depends on the broken feature. (status.openai.com) ### Why do file uploads matter so much? Because file upload has quietly become one of the most important ways people use ChatGPT. A lot of work now starts with “here’s the PDF,” “here’s the CSV,” or “look at this screenshot and tell me what’s wrong.” If that step fails, the model may still be online, but the useful part of the session is gone. Basically, the chat box is there, but the job can’t start. (status.openai.com) That is especially painful for analysts, students, and teams using ChatGPT as a document-reading layer. ### Why is Codex Cloud in the same incident? That’s a clue that the problem may have touched shared backend systems rather than just one front-end button. OpenAI’s notice grouped ChatGPT uploads and Codex Cloud task creation into the same event, which usually points to some common infrastructure path — storage, job orchestration, or another service both products rely on. That part is an inference, not something OpenAI spelled out in the status note. (status.openai.com) ### Is this unusual for OpenAI? Not really. OpenAI’s public history shows recurring incidents around file uploads, file processing, conversation errors, and access issues across ChatGPT and related products. That does not mean the platform is unreliable overall — the current status page shows ChatGPT at 99.82% uptime for the recent period — but it does show where the service tends to get fragile as more features pile on top of the basic chat interface. (status.openai.com) ### What should users take from this? The big lesson is that ChatGPT is now infrastructure for a lot of people, not just a novelty tab in the browser. When uploads or cloud coding tasks fail for a few hours, it interrupts real deadlines. OpenAI fixed this one fairly quickly. But the incident is a reminder to keep backup paths handy — local tools, saved prompts, and a way to keep moving when the AI layer stumbles. (status.openai.com) ### Bottom line OpenAI got ChatGPT and Codex Cloud back to normal the same night. But this outage showed where the platform is most load-bearing now — not plain chat, but the file and task features people build work around. (status.openai.com)