AI & Technology
What counts as an improvement in Codex’s 28-day pledge?
Tibo (@thsottiaux), whose profile identifies his work as Codex and ChatGPT at OpenAI, promised that each of the next 28 days would bring either a clear improvement relevant to most Codex/Work users or a full reset. Users immediately asked what qualifies and who decides.
The pledge followed a discussion about complexity. On October 4, Tibo described work on simplification, efficiency to support more usage, groundbreaking features and new models. He acknowledged that users wanted things to be simpler. Early on October 5 in Beijing, he quoted that response and added the 28-day commitment.
The condition was a clear improvement relevant to most users, rather than simply a release. Codex is an AI tool for programming tasks; Work is the product term used in his post. Baoyu summarized the pledge as daily releases or a usage reset. Gorden Sun questioned who would judge usefulness, whether one feature might be split across several days, and said he would prefer improvements to usage allowances and models. The pledge did not specify which allowances or accounts a full reset would cover, or when it would happen.
There was already a reset issue in the background. On October 3, Tibo said some subscribers he called “Pro 500” had not received the expected reset and that the team was investigating. He later reported a fix. Those posts establish his account of the investigation and repair; they do not independently confirm the outcome for every account.
The disagreement concerns whether a release translates into a practical benefit. Announcements can appear every day, while restored allowances and resolved problems require checking actual results. This issue records the pledge, responses and Tibo’s reported fix. The 28-day outcome is not yet available.