orywireNews that matters in AmericaReporting on public life, technology, money, and policy in the United States.
orywireOpenAI’s status page showed ChatGPT fully operational at the latest check, after June 3 incidents involving ChatGPT, Codex and the Responses API were marked recovered or resolved.
ChatGPT was not listed as down during the latest check of OpenAI’s status page. OpenAI said its systems were fully operational after several June 3 incidents involving ChatGPT, Codex and the Responses API were marked recovered or resolved.
OpenAI’s main status page said it was fully operational and that the company was not aware of issues affecting its systems. The page listed ChatGPT, APIs, Codex and FedRAMP among the service areas shown on the dashboard, with ChatGPT reporting 99.83% uptime for the March-to-June status period displayed there.
User-specific problems can still happen when an official outage is not active. People seeing errors should refresh the page or app, check their connection, sign out and back in, or check OpenAI’s status page before assuming a new outage is underway.
OpenAI’s history page showed three recovered June 3 incidents. The largest listed incident was “Elevated error rates on Codex, ChatGPT and Responses API,” which OpenAI classified as a partial outage and later marked resolved after saying all impacted services had fully recovered.
Earlier June 3 entries on OpenAI’s history page also showed “Elevated errors for ChatGPT Pro” and a Codex image-model error incident. Both were marked resolved or recovered.
The main June 3 partial outage listed affected components under APIs, ChatGPT and Codex. OpenAI’s incident page said one API component, two ChatGPT components and two Codex components were affected.
The separate ChatGPT Pro incident listed 12 affected ChatGPT components. OpenAI said some ChatGPT Pro users could see “Thinking Failed” when using the Pro model.
The Codex image-model incident listed five affected Codex components. It was not listed as a ChatGPT-wide outage.
For users still seeing errors, the most useful first check is OpenAI’s status page. If OpenAI shows no active incident, the problem may be account-specific, network-related, app-specific or temporary.
Basic steps include reopening ChatGPT, updating the mobile app, trying another browser, disabling a VPN if one is in use, and checking whether the same account works on another device. Developers should also check the API status and their own request logs before treating an error as a platform outage.
OpenAI’s public incident pages did not say how many users were affected by the June 3 disruptions. They also did not provide a detailed public cause for the main ChatGPT, Codex and Responses API incident.
This story should be refreshed if OpenAI posts a new active incident, adds a post-incident explanation or changes ChatGPT’s current operating status.
