Codex usage guides
Start with the question you have now. Every guide separates documented facts, practical interpretation, and account-specific details you still need to verify.
Understand the limits
How windows, models, and shared allowance behave.
How the Codex 5-hour limit resets
Understand the short Codex usage window, find its real reset time, and avoid starting the wrong countdown.
Read guide RESET NOTE / 03Codex banked resets, explained
Learn how a redeemable banked reset differs from recurring quota windows and public reset events.
Read guide USAGE RULE / 07Do Codex local and cloud tasks share limits?
Learn how local Codex messages and cloud tasks use the same included plan allowance and where to verify your current remaining usage.
Read guide USAGE RULE / 08Can switching Codex models save usage?
See when a smaller Codex model can make included usage last longer, how to switch with the CLI, and what switching cannot guarantee.
Read guide WINDOW NOTE / 02When does the Codex weekly limit reset?
Find your account-specific weekly recovery time and plan work without assuming a universal Monday reset.
Read guide USAGE RULE / 05What counts toward Codex usage?
Learn why model choice, task complexity, context, reasoning, tools, retrieval, and caching can change how quickly Codex usage is consumed.
Read guide USAGE RULE / 06Why does Codex show multiple usage limits?
Understand separate five-hour, weekly, and model-specific Codex meters without combining unrelated percentages or reset times.
Read guideTake the next step
Check, preserve, and stretch the quota available to you.
How to check remaining Codex usage
Find the five-hour and weekly meters, understand the percentages, and create a private reset countdown.
Read guide HOW TO / 09What to do when you reach a Codex usage limit
Check which Codex meter was exhausted, preserve active work, choose a documented continuation option, and set the correct reset reminder.
Read guide HOW TO / 10How to make Codex usage last longer
Apply OpenAI's documented ways to reduce unnecessary context and use a better-matched model before your next five-hour or weekly reset.
Read guide