OpenAI Outage Failed Up to 75 Percent of Requests After Database Maintenance Crashed Both Auth Replicas
A routine database maintenance job turned into one of OpenAI's worst outages of the year on Saturday, November 21, knocking out much of ChatGPT and the company's developer API for three and a half hours.
Between 1:48 and 5:21 p.m. Pacific time, request failure rates peaked at 75 percent across both paid and free ChatGPT and the API, according to a post-incident write-up OpenAI published on its status page the same day. The incident lasted 211 minutes in total, and all systems were reported operational again by Sunday morning, November 22.