OpenAI said on Wednesday it had applied a mitigation after users experienced elevated error rates for its GPT-4o-mini model in the application programming interface (API), and was monitoring recovery.
The company said the issue, which affects developers and businesses integrating its artificial intelligence tools via APIs, had been ongoing for about 10 minutes at the time of its status update.
"Users may experience elevated error rate for GPT-4o-mini in the API," OpenAI said, adding that mitigation steps had been implemented.
It did not disclose the cause of the disruption or give a timeline for full resolution.
The incident may have affected services that rely on the GPT-4o-mini model.
According to outage-tracking website Downdetector, the first incident was reported at 10:28 am, with a spike in reports observed around 10:43 am.
About 63% of users reported problems with ChatGPT, while 11% flagged issues with Codex, the website showed.
OpenAI did not specify the scale of the impact.
This is a developing story… please check back for updates