OpenAI Codex Lead Promises 28 Days of Daily Improvements or Quota Resets

Sottiaux described an improvement in broad terms. It could be a new release, or it could be a basic bug fix. The article said this either-or formula has become a regular approach for him. For users, either a new feature or a quota reset is generally welcome. But frequent resets also raise questions. Users may not use up their quota, and repeated resets can suggest that OpenAI is eager to please or lacks confidence in its product. The QbitAI article also asked whether pressure from Claude Opus 5.5 was too great.

Before the 28-day pledge, Sottiaux had been collecting feedback on what Codex still lacked and what users had long wanted but not received. After reviewing the responses, he narrowed the work to four areas: simplifying the product, improving efficiency to support more usage, breakthrough features, and new models.

During that feedback push, Lauren Tan of Grok Bot joined the conversation. She mocked OpenAI for shipping what she called slop and for relying on quota resets to carry the show. She also accused the company of maxing out publicity while quality suffers. Sottiaux responded in effect that she should do better herself if she could. The exchange had the flavor of mutual attention-seeking, but it also reflected dissatisfaction that goes beyond quota limits.

User complaints, as summarized by QbitAI, center on three issues. The first is focus. DevDay introduced more than 20 updates at once, making it difficult to tell what mattered most. The one item presented as a highlight, a personal agent called Dot, currently looks weak. The second is planning. Resets are meant to serve work usage, so that people do not feel quota running out while doing real work. But frequent and unpredictable resets make it hard to schedule work in advance. Quota plans are also shrinking, and a new $500 option is too expensive. The third is model reliability. Tasks disconnect midway, and no number of resets fixes a session that is stuck. The article compared this to having unlimited bullets but a gun that jams on every shot.

The key part of Sottiaux's promise, therefore, is not the daily rhythm but the phrase about being clearly useful to most users. Fewer disconnects, less rework, and less time spent studying an entry point all count as improvements. A major release is not required. After 28 days, the best outcome for Sottiaux may be that users pay less attention to his updates.

Editor's Summary

OpenAI Codex lead Tibo Sottiaux has promised 28 days of daily improvements or full quota resets for Codex and Work users, according to QbitAI. The pledge follows complaints about unclear product priorities, shrinking and unpredictable quotas, and model reliability, as well as a public exchange with Lauren Tan of Grok Bot. The move may ease short-term frustration, but it does not by itself address the underlying product issues users raised.