AI TRENDS | OpenAI Codex Team Member Says Update Caused Confusion Over Usage Limits and Desktop Features

OpenAI has rolled out initial fixes after a recent update led to user confusion about high-compute settings, made common desktop app features harder to find, and caused issues with some multi-agent workflows and plugins. According to Odaily, OpenAI Codex core team member Tibo wrote on X that users were previously able to enable high-compute mode too easily without clear explanation of how it would affect usage quotas. He also said a major redesign of the desktop app made frequently used functions such as chat and projects difficult to locate, and that promotional messaging focused heavily on ChatGPT Work led some Codex users to believe Codex might be phased out. Tibo added that the new version introduced problems affecting some multi-agent workflows, plugin functionality, and other aspects of the user experience. OpenAI’s first set of improvements includes adjusting default settings and the model selector to reduce accidental use of higher-cost configurations, fixing multiple plugin submission issues, improving how Codex is presented within the product, and addressing some desktop experience problems. OpenAI said it has spent the past 24 hours collecting user feedback, analyzing usage, and communicating with users. The company also acknowledged multiple issues with the update. OpenAI has completed two usage-quota top-ups for Codex and ChatGPT Work so far. It said a larger update is scheduled for next week, with chat and project functions returning to the sidebar in a more familiar and customizable format. The update will also increase visibility into usage quotas and reset times and further clarify the respective use cases for ChatGPT Work and Codex.