OpenAI Fixes Three Quality Issues in Codex Astria Ahead of Usage Reset

OpenAI has fixed three quality issues confirmed in Codex Astria and plans to reset user usage ahead of midnight local time on the 12th. Codex product lead Thibault Sottiaux stated that fixes were made for issues where skills intended for previous Astria models were over-executing or blocking model task reviews. Experimental context management features, which caused tasks to terminate prematurely or respond to prior messages, have been disabled, affecting an estimated 4,000-5,000 users. Sottiaux also mentioned that a configuration error engine causing quality degradation for some users has been removed.
Korean Source
This article is an English localization of a Korean-language crypto news report. Original headline: 오픈AI, 아스트라 품질 문제 3건 수정…사용량 재설정 예정