Codex Weekly Limit: How the 7-Day Quota Works & Resets
The Codex weekly limit is the 7-day cumulative usage ceiling for your OpenAI account. It bounds total model usage across the Codex CLI, IDE extension, desktop app, and cloud background tasks (sharing a unified usage pool with ChatGPT Work). It operates alongside your five-hour usage window: you must have allowance remaining in both to submit prompts. OpenAI does not publish an exact weekly token cap, but publishes relative plan tiers and official reset reanchoring rules for paid resets.
Caps overall model consumption across a seven-day cycle. If this allowance is exhausted, prompt turns pause until your account's scheduled weekly reset anchor or until an eligible reset is applied.
Meters short-term usage across a five-hour window. If this limit is reached before five hours pass, prompting pauses until the five-hour window ends, without necessarily exhausting your overarching weekly allowance.
Does OpenAI Publish an Exact Weekly Allowance?
No. OpenAI does not publish a fixed token number, message quota, or dollar cap for the weekly limit on any subscription tier.
Third-party claims citing exact weekly token quotas (such as “5 million tokens per week”) are unverified assumptions. OpenAI's current public documentation does not publish an exact weekly token or message ceiling. Instead, the amount of included usage consumed by a task varies based on documented factors including the model selected, reasoning settings, inputs and outputs, and whether tasks involve multi-step execution.
- Published: Relative usage tiers — Pro $100 includes 5x Plus usage, Pro $200 includes 20x Plus usage, and Business Premium includes 5x more included usage than Business Standard.
- Published: Estimated 5-hour local message ranges for supported models (e.g., 5–45 messages for GPT-6 Astra on Plus; 250–2,000 for GPT-5.6 Luna).
- Undisclosed: Exact weekly token ceilings or message counts for the weekly allowance.
How to Check Your Remaining Weekly Quota and Reset Time
OpenAI reports your live weekly status through three direct surfaces. Each surface displays your remaining percentage and the exact date and timestamp when your weekly quota refills:
Run /status at any time inside an active Codex CLI session. Codex prints your 5-hour remaining percentage, weekly remaining percentage, and the next reset time in UTC.
In the ChatGPT desktop app or web browser, navigate to Settings > Usage (or visit chatgpt.com/codex/settings/usage). This panel displays your active billing period, visual progress bars for both windows, and your account-specific reset anchor.
Install Codex Usage, our free local tool. It reads your authenticated session locally to display a native macOS menu bar meter, terminal readout, or local web dashboard that tracks countdowns and flags whether your 5-hour or weekly window is under greater pressure.
Why You Can Have 5-Hour Allowance Left but Still Be Blocked
Because the 5-hour and weekly limits track concurrent meters, your ability to prompt Codex requires that neither meter is at 0%. Developers frequently encounter two distinct lockout scenarios:
Why it happens: You paced your work throughout the week without triggering short-term rate limits, but your aggregate prompt turns hit your 7-day quota ceiling. Even though your 5-hour meter is fresh, you cannot prompt until your weekly reset arrives.
Why it happens: You submitted tasks with large context, ran multi-step tasks, or used GPT-6 Astra with higher reasoning settings in an active session. Your 7-day budget may still have capacity, but you reached the five-hour limit and must wait for the five-hour window to end.
What Happens When Your Weekly Allowance Is Exhausted
When your weekly quota hits 0%, Codex executes a controlled pause rather than terminating abruptly:
OpenAI officially documents that Codex can continue working on the turn already in flight when the limit is reached, subject to fair-use limits. Your active code generation is not discarded mid-stream.
Once the active turn finishes, submitting additional prompts in the CLI, IDE, or desktop app will trigger a rate-limit notice showing the scheduled weekly reset time.
Your local Git branches, staged files, terminal history, and session context remain completely intact. You never lose local code when hitting subscription limits.
To unblock your workflow when your weekly allowance is exhausted, you have four distinct paths:
- Wait for your natural weekly reset: Check
/statusor Settings > Usage for your account's recovery timestamp. - Purchase a paid instant reset: Eligible Plus and Pro users can purchase an immediate refill in Settings > Usage to restore applicable five-hour and weekly allowances (see reanchoring rules below).
- Redeem a banked reset: Apply an eligible promotional grant stored in your account to refresh your five-hour and weekly Codex windows (see our Codex banked resets guide).
- Switch to API key billing or credits: Switch your CLI to an OpenAI Platform API key to pay metered rates per token, bypassing subscription windows entirely (see Codex credits & API billing).
Weekly Limit Differences Across Plans
How weekly allowance behavior varies between consumer plans and workspace seats, based strictly on official documentation:
| Plan Tier | Weekly Allowance | Window Mechanics | Reset & Recovery Options |
|---|---|---|---|
| Plus ($20/mo) | Baseline plan usage (exact weekly cap not disclosed) | Dual window: five-hour limit + weekly quota | Natural weekly reset, paid instant reset, banked reset, credits |
| Pro $100 ($100/mo) | 5x usage tier (5x more usage than Plus; exact weekly cap not disclosed) | Dual window: five-hour limit + weekly quota | Natural weekly reset, paid instant reset, banked reset, credits |
| Pro $200 ($200/mo) | 20x usage tier (20x more usage than Plus; exact weekly cap not disclosed) | Dual window: five-hour limit + weekly quota | Natural weekly reset, paid instant reset, banked reset, credits |
| Business Standard ($20–$25/mo/seat) | Baseline included usage per seat (exact weekly cap not disclosed) | Dual window: five-hour limit + weekly quota | Natural weekly reset or workspace credit pool (no personal instant resets) |
| Business Premium ($100–$125/mo/seat) | 5x more included usage than Standard (exact cap not disclosed) | Weekly window only (no five-hour limit on Premium) | Predictable weekly reset or workspace credit pool (no personal instant resets) |
For the complete comparative pricing breakdown and cost economics, see our OpenAI Codex pricing guide and general Codex limits table.
Paid Instant Resets & The 7-Day Reanchoring Rule
For eligible personal ChatGPT Plus and Pro accounts, OpenAI offers paid weekly limit resets. The mechanics of this feature are governed by explicit rules defined in OpenAI Help Article 20001507:
The First-Request Reanchoring Rule
When you purchase a paid instant reset, your weekly usage allowance is pulled forward immediately. However, your new 7-day billing clock does not begin at the moment of payment:
If you purchase a reset late Friday night but do not send another prompt until Monday morning at 9:00 AM, your next 7-day weekly reset anchors to the following Monday at 9:00 AM. You do not forfeit quota during idle weekends.
Paid instant resets apply immediately upon checkout and cannot be stored in reserve for later use; their 7-day period begins on your first request post-reset. For promotional banked resets, OpenAI documents that applying a full banked reset refreshes both the five-hour and weekly Codex windows and changes your weekly reset date (visible in Settings > Usage). Banked resets can be held until activated or expired (see our Codex banked resets guide).
Frequently Asked Questions About Codex Weekly Limits
Does OpenAI publish the exact token or message number for the weekly limit?
No. OpenAI does not publish an exact token or message ceiling for the weekly allowance. It officially publishes relative tiers between plans (such as Pro $100 offering 5x Plus usage, Pro $200 offering 20x Plus usage, and Business Premium offering 5x more included usage than Standard) and publishes directional 5-hour message estimates by model.
Why can I have 5-hour quota remaining but still be blocked from prompting Codex?
Codex meters two concurrent limits: a short-term five-hour usage window and an overarching seven-day weekly window. If your weekly allowance is exhausted, you are blocked from submitting new prompts even if your five-hour window shows available capacity. To continue prompting, you must have available allowance in both windows simultaneously.
Does buying an instant reset give me extra quota beyond my regular weekly limit?
No. A paid instant reset does not grant bonus tokens or increase your plan's maximum capacity. As documented in OpenAI Help Article 20001507, purchasing a reset immediately pulls your normal weekly allowance forward, restoring the applicable five-hour and weekly allowances for your plan.
How does purchasing an instant reset reanchor my weekly schedule?
When you purchase an instant reset, OpenAI documents that your new weekly usage period begins with your first request in Work or Codex after the reset is applied. Your next automatic weekly reset is then scheduled 7 days after that first request. (For promotional banked resets, OpenAI documents that applying a full banked reset refreshes both the five-hour and weekly Codex windows and changes your weekly reset date, which you can verify in Settings > Usage).
What happens to a command or task currently running when my weekly allowance hits zero?
Codex does not abruptly terminate an active execution turn. According to official documentation, the model can continue working on that in-flight turn subject to fair-use limits. Once that turn finishes, subsequent prompt submissions pause until your scheduled weekly reset arrives or you use an available reset option.
Are weekly limit resets available for ChatGPT Business or Enterprise workspaces?
No. Paid instant resets are restricted to personal Plus and Pro accounts. Team workspaces on ChatGPT Business, Enterprise, and Edu manage capacity caps through seat tier upgrades (such as Business Premium at $100/seat/month billed annually or $125 billed monthly, which includes 5x more usage than Standard and removes the five-hour limit) or by enabling shared workspace credit pools in billing settings.
The facts and reanchoring mechanics on this page are verified against official OpenAI documentation:
- Codex 5-hour limit guide — why it isn't 5 hours of wall-clock coding, how a new window starts, and published message estimates.
- when does Codex usage reset — the broad reset schedule across both meters and global refills.
- Codex instant resets — pricing, eligibility, and desktop checkout.
- Codex banked resets — promotional reset benefits and how to redeem them.
- Codex usage limits table — full plan comparison across Free, Plus, Pro, Business, and Enterprise.
- ChatGPT Plus Codex limits — the $20/month tier in detail.
- ChatGPT Pro Codex limits — 5x ($100/mo) and 20x ($200/mo) tiers.
- ChatGPT Business Codex limits — Standard vs. Premium seats and the 5-hour limit removal.
- Codex credits & API billing — overflow capacity when included allowances are depleted.