OpenAI has reset usage quotas for Codex and ChatGPT Work following user confusion over high compute mode settings. Tibo, a core member of OpenAI Codex, revealed that users were inadvertently enabling high compute mode without understanding its impact on usage quotas, affecting multi-agent workflows and plugin features. To address these issues, OpenAI has adjusted default settings and the model selector to prevent accidental high-cost configurations. The company has reset usage quotas twice and plans further updates next week. These updates will include the return of chat and project features to the sidebar in a more customizable form, and improved visibility of high compute mode usage quotas and reset times. These changes aim to clarify the scenarios applicable to ChatGPT Work and Codex, ensuring users have a better understanding of their usage limits.