OpenAI’s status page listed an active incident on Thursday afternoon under the title “Elevated errors across ChatGPT and Codex”, with the status set to monitoring and the note: “We have applied the mitigation and are monitoring the recovery.”
What users saw
Reports describe a 404 rather than a timeout — the page returning “This chatgpt.com page can’t be found”, and the API surfacing “unexpected status 404 Not Found: Unknown error”.
A 404 is an odd failure mode for an outage. It says the route was not found rather than that the service was overloaded, which usually points at routing or configuration rather than capacity.
Scope
OpenAI’s own status entry names two components: ChatGPT and Codex. Third-party trackers logged a spike in reports through the afternoon, with users citing both the web interface and the mobile apps.
Aivio News has not independently verified the tracker figures, which are self-reported by users rather than measured.
Where it stands
At the time of writing the incident was still open, marked as monitoring after a mitigation was applied. That status means service is expected to be recovering, not that the incident is closed.
OpenAI had not published a cause. Status-page entries during an incident rarely include one; a post-incident note, if it comes, usually follows days later.
What to watch
Whether OpenAI publishes a root cause. Codex failing alongside ChatGPT points at shared infrastructure rather than a single product, and that distinction matters to anyone whose build pipeline depends on one of them.
