OpenAI's Codex lead promises a fix or usage reset daily for 28 days
Tibo Sottiaux, who leads Codex at OpenAI, said each day for 28 days will bring either a clear improvement or a full usage reset. The post drew more than 9,000 likes within two hours.
Tibo Sottiaux, who leads Codex at OpenAI, said on X that over the next 28 days the team will each day either ship one clear improvement relevant to most Codex and work users, or ship a full reset. The post had about 9,280 likes and 1,437 replies when read on Sunday evening UTC.
https://x.com/thsottiaux/status/2106845241357824205
A reset appears to restore a user's weekly usage allowance, and the practice is familiar to Codex users. Sottiaux has pressed it often: the developer Theo wrote on Sunday that, as he remembers July, OpenAI subscribers got resets from Sottiaux about every three days on average over the summer.
OpenAI has not published a changelog for the programme or said what the daily improvements will be. TestingCatalog replied that he did not know whether testing 28 new things or 28 resets was better, adding that it looked busy. Sottiaux has not said how a reset might be chosen over a feature on a given day, and the programme could change.
Reaction was mixed. The commentator Kimmonismus said the plan felt like an attempt to stop users defecting to Claude after DevDay shipped few releases he thought justified the hype, and said subscription plans had effectively been cut in half. Those are his views and figures, and OpenAI has not confirmed a change in plan limits.
The comparison with Anthropic is already live. Theo posted that Claude Opus 5.5 is now the stronger coding model and that Anthropic's $200 plan lasts longer than it did in July. He rated Opus 5.5 9 out of 10 for code against 7.5 for GPT-6 Astra. Those ratings are his own impressions, not a benchmark.