Technical blog · published

When Codex Goes Down, the Reply Matters

Ryan Spice 3 min read

Tibo acknowledged a Codex outage, later said service was back, and announced a usage reset for paid Codex and ChatGPT Work users. The quick updates helped close the loop.

Stylized illustration of a hooded person holding a mug beside a puppy and a supportive sign addressed to Tibo.
Illustration shared by Ryan Spice via X
Contents3 sections
Back RSS

Codex went down on September 25, and Tibo acknowledged it publicly. A lighthearted reply followed about ten minutes later. Later that evening, he said service was back and announced a usage reset for paid Codex and ChatGPT Work users. The speed of that sequence matters when people have stopped work and are waiting to know what happens next.

A quick acknowledgment removes one uncertainty

An outage creates two problems for a user. The tool is unavailable, and the user does not know whether the failure is local, account-specific or widespread. A clear public acknowledgment answers the second question immediately. I can stop debugging my setup, avoid repeating failed requests and decide what work can continue elsewhere.

That is why turnaround matters even before a fix exists. A short message from a recognizable product lead gives the team a visible point of accountability. It also gives users somewhere to look for the next update. The message does not need a polished root-cause analysis while the incident is still unfolding. It needs to be timely and accurate.

His follow-up was simply “o no :(”. That small joke works because the serious acknowledgment came first. Humor can make a difficult moment feel human; it cannot substitute for status, recovery and a later explanation if the impact warrants one.

A promised reset can make recovery tangible

For a usage-limited coding tool, an outage can cost more than minutes. A failed or interrupted session can break concentration and leave users unsure how much of their allowance the attempt consumed. A usage reset gives people room to restart and recognizes the interruption on the user's side of the screen.

Tibo's recovery post says the team is back in action and will reset usage limits for all paid users across Codex and ChatGPT Work. That is an announced commitment, not account-level proof that the reset has already appeared for every eligible user. The post does, however, answer the question the outage left open: who is included and what the team plans to do next.

The sequence is simple: acknowledge the problem quickly, keep the status current, restore the product, then give users a clean way back into their work. A reset does not replace reliability. Paired with an honest recovery update, it can turn a frustrating interruption into evidence that the team understands the cost of lost momentum.

What I verified

I checked the three posts in Tibo's X thread on September 25: the outage acknowledgment, the “o no :(” reply, and the recovery and reset announcement. The last post promises a reset for paid Codex and ChatGPT Work users; I did not verify its account-by-account rollout or measure the outage's impact. OpenAI's status history is another place to check for an official incident record. The accompanying image is an illustration, not a photograph of Tibo or evidence of the outage.

Published
Updated
Author
Ryan Spice

Sources

Sources

Primary documentation and source material used for the factual claims in this article.

Back RSS