In the early morning of December 27, OpenAI announced,The chatbot ChatGPT, video generation model Sora and API have experienced major errors. Most of them have been down for several hours, but the problem has been identified and recovery has begun. The outage began at around 1:30 pm Eastern Time on December 26. Network status monitoring website Downdetector stated that in the afternoon, more than 15,000 users reported problems, mainly related to ChatGPT. By the evening of the same day, the number of reports had dropped to less than 700.


OpenAI immediately issued a report stating that ChatGPT, API and Sora currently have high error rates, and the problem is caused by upstream providers.


Image source: OpenAI

According to previous media reports, OpenAI did not specify the "upstream provider" related to the problem, but its exclusive cloud provider Microsoft reported that there was a "power problem" in one of its data centers. The problem occurred at the same time as the OpenAI problem and affected North America. At the same time, there were also problems with Xbox cloud games. Just after 5 p.m. ET on December 26, Microsoft said it had "fully restored power" to the affected data center.


As of the end of this summer, ChatGPT has more than 200 million daily active users. Since its release, popular OpenAI products including ChatGPT have experienced multiple outages.

The most recent large-scale outage occurred on December 11, a few days after the release of Sora. All services under OpenAI, including ChatGPT, API, and Sora, experienced severe performance degradation or even complete unavailability from 3:16 pm to 7:38 pm Pacific time on December 11, lasting more than four hours. This outage was due to misconfiguration of the newly deployed telemetry service, which overloaded the control planes of hundreds of Kubernetes clusters around the world, causing cascading failures of critical systems.