OpenAI said some users were facing a ChatGPT outage with elevated error rates on ChatGPT and Codex, and it had already applied a mitigation while watching recovery. That left a partial service problem in place for affected users, not a clean fix. Marcus Mendes tracked the update as the incident stretched on.
OpenAI status page
OpenAI’s system status page said, “We are investigating the issue for the listed services.” The same update read, “Investigating · Ongoing for 4 hours · Affects ChatGPT, Codex.”
Those four hours matter because they point to a live service issue that had not settled when the update went out. For anyone depending on either product, the practical signal was simple: retrying might work, but the service was still unstable.
Marcus Mendes and X
Marcus Mendes, a tech podcaster and journalist at 9to5Mac, said OpenAI had applied a mitigation around two hours earlier and was monitoring recovery. He also noted that reports on DownDetector had trended down after sharp spikes over the past few hours.
That combination suggests the problem was easing, but not gone. Users on X still reported the issue as ongoing, which means some people were still running into errors even after OpenAI changed course on the back end.
ChatGPT and Codex recovery
The incident affected ChatGPT and Codex at the same time. For users, that means the disruption was not limited to one interface or one workflow, and it touched both a consumer-facing service and a developer tool.
OpenAI moves GPT-5.6 to general availability amid Chatgpt Outage price race sits in a different lane, but today’s outage is the more immediate test: whether OpenAI can keep both services stable while the mitigation keeps working.
The open question is what caused the elevated error rates in the first place. OpenAI has said it was investigating the issue, but the status update stopped at mitigation and recovery monitoring.
