Will Codex Reset Today?
Verified Codex resets · 24–48h forecast
Did Codex reset today? No — not so far today (UTC). The latest verified global reset was 10 days ago (Sep 12, 8:09 AM UTC).
last global reset
10 days ago
- 40 verified resets
- pace ≈ every 6d
- longest wait 73d
❯ will codex reset? yes — Tibo committed to a reset signals
⚠️Ladies and gentlemen... start... your... ENGINES. We are almost Tuesday and I promised a reset for Tuesday. Among some other things. See you soon.latest reset alert · end of Tuesday · 4 hours agoView on X
Will you make it to your next reset?
how it works
Reference account first; your `/status` time overrides it. Add your usage % to compare pace — all data stays in this browser.
❯ codex /status
❯
enter the exact time from /status · slide in the usage from /status · paste /status instead · / Command palette
FORECAST NOTES
Global reset outlook, without black-box scoring
Codex reset calendar
see the reset rhythm chart- verified reset
- last global reset
- banked reset update· not a reset · never moves the clock
All-time median gap: 4.2 days · last 30 days: 5.5 days · tracked since Sep 2025
Goodwill resets verified from @thsottiaux posts · a rhythm, not a schedule.
what is this?
This model tracks irregular goodwill resets for everyone. Your personal burn meter above stays private and account-specific.
Experimental · not above baseline
The homepage answers one question first: when could a new reset signal appear? Waiting longer does not raise the odds — models that treated a reset as becoming "due" scored worse in the backtest, so the 24h number stays flat while the quiet stretch runs.
- Recent cadence: a reset every ~2.2d across the last 5 (weighted mean 5.7d) Details
- Last reset: 2026-08-31 (4.4d ago) Source
- Most common window: 11 PM - 2 AM in UTC Details
open verified timeline · see the reset rhythm chart · open full Tibo radar
RADAR
Did Codex reset today?
Tibo's latest posts, classified live into reset, banked and limits signals, next to the verified reset log and the hour-of-day pattern.
Ladies and gentlemen... start... your... ENGINES. We are almost Tuesday and I promised a reset for Tuesday. Among some other things. See you soon.
2.5K98017.3KOK fine. But it’s also still coming in Tuesday
4631877.2K@stemonteduro You main Sol? And out of curiosity, you never buy credits?
1.2K161.4KCan’t believe this was filmed this weekend, neck tattoos and all
533998Reset all propagated. Sweet dreams. t.co/VgKVUixoJG
2.9K77117.6KHi Astra users. A reset and a quick update on quality issues that have been posted around. Working with some of you, we have found and fixed the following issues: - Some skills written for previous models were triggering too often or preventing the model from checking its work. - An opt-in context management experiment that could cause early stops or replies to older messages. We've disabled it. Our rough estimate is that 4-5k users were affected by this experiment. - We've also removed some badly configured engines that resulted in a measured quality degradation for a long tail of traffic flowing through them. We’ve also made some more minor improvements and things should feel significantly better across the board. More consistent follow-through, better tracking of your latest message, and better checks on the work as it’s going through the motions. The examples posted and all the users who worked directly with us were incredibly useful in helping fix things quickly. Always grateful for this incredible community. And of course, a reset is also landing by midnight today.
4K1.6K28.5KNext week we’ll be retiring GPT-5.3-Codex-Spark. Can you believe we shipped a model named as such!! It's had a good run and was a lot of fun, but usage has been declining and we have significantly better models now. Time to make room for the future.
1.9K39615.2KWhen I say excellent service for existing users, that includes the occasional reset
9161467.5K@aubreystrobel Just checked and we do have a few weekly Codex users in the Vatican now
102272KTo make sure our current users have an incredible experience and continued access to Astra, we are going to pause subscriptions to our $200 Pro plan. These put the most strain on our systems and we wanted to take the smallest step that allows us to continue giving the broadest access possible. All other plans and the api remain available. There is no impact to existing accounts and we are working on adding more capacity as fast as we can. Thanks!
2.9K1K17KThere was a bit of a kerfuffle this morning with some banked resets not fully applying when used in ChatGPT Work and Codex. Everyone who used one in the affected time window is getting another one and an email to apologize.
2.2K25110.2KI am pleased to see that Ant’s new Claude Code has a version of background computer use on par with the version from Codex from last May this year. Nice work! Shipping great features first turns out to be a great way to encourage other labs to ship too. We will continue to do this until other labs pay more attention to shipping. This is good for everyone and there is a lot of room left to go! We solved computer use in practice for GPT models about four months ago. But computer use is a significantly awesome thing no matter what model you use, and it is important that the industry similarly spends more effort to train their models to be great at computer use, among other elements of model capabilities. As models become more capable and central to businesses and economies, the value of computer use only increases. We should be taking it seriously, and doing the right thing for our customers and the world.
4131796.1K
recorded global limit events
- 2026-09-22RESETLadies and gentlemen... start... your... ENGINES. We are almost Tuesday and I promised a reset for Tuesday. Among some other things. See you soon.
- 2026-09-12RESETReset all propagated. Sweet dreams. https://t.co/VgKVUixoJG
- 2026-09-12RESETHi Astra users. A reset and a quick update on quality issues that have been posted around. Working with some of you, we have found and fixed the following issues: - Some skills wr
- 2026-09-09BANKED · CHECK USAGE RESETSThere was a bit of a kerfuffle this morning with some banked resets not fully applying when used in ChatGPT Work and Codex. Everyone who used one in the affected time window is get
- 2026-09-08RESETNever gonna give you up Never gonna let you down Never gonna run around and desert you Never gonna make you cry Never gonna say goodbye Never gonna tell a lie and hurt you Thanks
- 2026-09-03BANKED · CHECK USAGE RESETSAstra access compensation: one banked reset for every day without Astra on a paid ChatGPT plan, starting that day — the first drop promised about three hours later.
- 2026-08-31RESET25M active users: to celebrate, usage reset for every paid ChatGPT Work and Codex subscription — the reset promised for 6pm PT the previous day.
- 2026-08-25RESETQuiet reset for all paid plans — weekly usage back to 100% from ~14:00 UTC. Tibo confirmed it only in replies ("Ah yeah, forgot to say") ahead of the Plus 5h-limit return.
avg 7d between recorded events
What time of day do Codex resets happen?
Every recorded announcement, bucketed by hour in your timezone. Resets cluster in the US afternoon — if you're waiting on a goodwill reset, that's the window to watch.
55 announcements bucketed by hour in your timezone (UTC) · peak 23:00
LIMITS(5)
How Codex usage limits work
the 5-hour window
The short clock. Local messages (CLI, IDE extension, desktop app) and cloud chats draw from one shared 5-hour budget — they are not separate allowances. When your Codex CLI reports this window, /status shows its percentage left and reset time; use the rows your own client actually prints.
the weekly window
The ceiling behind it, running seven days. OpenAI has never published how large it is for any plan — the official pricing page says only that "additional weekly limits may apply". Clearing a 5-hour window does not restore weekly budget; the two refill independently.
what a reset actually is
The reset time is not computed on your machine. OpenAI's servers send an absolute resets_at timestamp with each usage report, and the CLI just renders it. A reset restores your percentage to 100% — it doesn't add credits or change your plan. Community reports suggest each window's clock starts on your first request after a reset, though OpenAI hasn't confirmed that.
goodwill & banked resets
Separately from scheduled windows, OpenAI may apply an automatic reset or save a banked reset for later. A banked grant does not refresh usage; redeeming a full banked reset resets both 5-hour and weekly windows. Referral resets normally use a 30-day window unless the offer says otherwise; every grant follows the expiry shown by its offer or account.
local messages per 5-hour window
| model | Plus $20 | Pro 5× $100 | Pro 20× $200 | Business |
|---|---|---|---|---|
| GPT-6 Astra | 5–45 | 25–225 | 100–900 | 5–45 |
| GPT-5.6 Sol | 10–100 | 50–500 | 200–2,000 | 10–100 |
| GPT-5.6 Terra | 25–200 | 125–1,000 | 500–4,000 | 25–200 |
| GPT-5.6 Luna | 250–2,000 | 1,250–10,000 | 5,000–40,000 | 250–2,000 |
| GPT-5.5 | 15–80 | 75–400 | 300–1,600 | 15–80 |
OpenAI's own published estimates, not fixed caps; real counts vary with task size, context, reasoning effort, tool use and caching. GPT-6 Astra gets the smallest window of any model — about half of Sol's. Business per-seat numbers match Plus, while the $100 Business seat uses the Pro 5× estimates; Pro tiers are exact 5× / 20× multiples. Cloud chats run GPT-5.6 Sol and can use more allowance than local messages, and Fast mode bills Astra at 2.5× its standard rate. GPT-5.4 and GPT-5.4 mini retired from Codex on Aug 31, 2026 (OpenAI maps them to Terra and Luna). Weekly caps are not published for any plan. Source: learn.chatgpt.com/docs/pricing. Verified 2026-09-06.
plan-level summary
| plan | codex access | when you hit the wall |
|---|---|---|
| Free / Go | included for quick, lightweight tasks — no published numbers | wait for the window to roll over |
| ChatGPT Plus $20 | GPT-6 Astra (all Plus users since Sep 4, 2026) plus the full GPT-5.6 family across CLI, IDE, web, cloud | buy credits (Settings → Usage → Credits) or wait |
| ChatGPT Pro $100 / $200 | 5× or 20× the Plus windows, plus GPT-6 Astra Pro and the Spark preview | credits with auto top-up, or wait |
| Business | Plus ($20 seat) or Pro 5× ($100 seat) per-seat windows + admin controls; Astra since Sep 4, 2026 | workspace credits |
| Enterprise / Edu | flexible pricing = no fixed rate limits; usage scales with credits | credits pool |
| OpenAI API key | pay per token — no 5h or weekly windows at all | only your own spend limit (loses cloud features) |
Codex, ChatGPT Work, ChatGPT for Excel and Workspace Agents share one agentic usage pool; ordinary ChatGPT chat is metered separately and never drains it. Astra draws on that same allowance on every paid plan — 100% of it can go to Astra — at 2.5× Sol's credit rate. OpenAI quotes GPT-5.6 usage at 5–30 credits per message; credit prices vary by account and region, and purchased credits expire after 12 months. Source: OpenAI Help Center: Using Codex with your ChatGPT plan. Verified 2026-09-06.
FAQ
Codex reset questions, answered
When does my Codex 5-hour limit reset?
Your Codex client shows the windows currently returned for your account. Run /status in the Codex CLI and read the percentage left and reset time from the limit-window rows it actually prints. If a 5-hour limit (quota) row appears, enter that row's reset time in the tracker above; if it does not, do not infer one from this page.
When does the Codex weekly limit reset?
Seven days on its own clock, shown in the "Weekly limit" row of /status for your weekly quota. There is no universal weekly reset moment shared by everyone: community reports consistently indicate each account's window anchors to its own first request after the previous reset. That's why this tracker asks for your reset time instead of pretending one global countdown fits all.
How big is the weekly limit on my plan?
OpenAI does not publish it. The official pricing page says only "additional weekly limits may apply" and gives no weekly number for any plan — anyone quoting an exact weekly cap is guessing. What is documented: the weekly window runs seven days and refills independently of the 5-hour window.
Did Codex reset today?
Check the radar above — it watches @thsottiaux (the Codex lead, who announces virtually every goodwill reset), flags rate limit reset announcements automatically, and shows how long ago the last one landed. OpenAI has granted 38 recorded global resets since September 2025, including 10 in the twelve days after GPT-5.6 launched, and the GPT-6 Astra rollout (Sep 3–5, 2026) added a banked reset for every day a paid account lacked Astra plus a full banked reset for all Plus, Pro and Business users once it reached everyone.
Does clearing my 5-hour window restore my weekly limit?
No. They are independent budgets refilling on separate schedules. If your weekly window is exhausted, a fresh 5-hour window won't unblock you — you need the weekly rollover, a goodwill reset, a banked reset, or credits.
Are Codex limits separate from my ChatGPT chat limits?
Yes, but Codex isn't alone in its pool: Codex, ChatGPT Work, ChatGPT for Excel and Workspace Agents all share one agentic usage pool. Ordinary ChatGPT chat has separate caps and never drains Codex budget — a heavy ChatGPT Work session will, though.
What is a banked reset, and when does mine expire?
A reset saved to your account until you choose to redeem it. The grant itself does not refresh usage; a full redemption resets both the 5-hour and weekly windows and moves the weekly date to roughly seven days later. Referral resets normally must be used within 30 days unless the offer says otherwise, while milestone, compensation, and one-time grants follow the expiry shown by the account or offer. Read the banked reset guide
What time of day do Codex resets usually happen?
From the 50 recorded announcements: they cluster between 17:00 and 23:59 UTC — US afternoon — with a secondary bump just after midnight UTC. The histogram above localizes every announcement to your timezone. Day of week is nearly uniform, so don't bother planning around one.
What happens when I hit a limit mid-task?
OpenAI lets the agent finish the turn in progress (subject to fair use), then blocks new turns until a window clears. Your session and history stay put. Options: wait it out, spend a banked reset, buy credits (Plus/Pro: Settings → Usage → Credits; valid for 12 months), switch to a lighter model like GPT-5.6 Luna (its 5-hour estimate is roughly 50× Astra's), or move the task to an API key.
Can I dodge subscription limits with an API key?
Yes, with a trade-off: signing Codex in with an API key bills per token at standard API rates and removes the 5-hour and weekly windows entirely — but you lose the cloud features (GitHub code review, Slack, Codex cloud), and your only ceiling is the spend limit you configure.
Is the 5-hour limit still lifted?
Not for Plus. OpenAI removed it for paid plans on July 12, 2026 during the GPT-5.6 demand surge, then Tibo announced on August 25 that the 5-hour limit would return for Plus accounts across ChatGPT Work and Codex the next day, to smooth compute load. No matching return has been announced for Pro or Business. The pricing page publishes 5-hour estimates for every model, GPT-6 Astra included — check your own /status for the live answer, and watch the radar either way.
Does GPT-6 Astra have its own usage limit?
No separate meter. Astra is included in the normal allowance on every paid plan and you can spend 100% of it on Astra, but it is the most expensive model in the pool: OpenAI estimates 5–45 local messages per 5-hour window on Plus, against 10–100 for Sol and 250–2,000 for Luna, and Fast mode bills Astra at 2.5× its standard rate. During the Sep 3–5, 2026 rollout OpenAI banked one reset per day for paid accounts without Astra, then a full banked reset for all Plus, Pro and Business users.
Can this site see my real Codex usage?
No — this site never touches your OpenAI account. Your real numbers live in OpenAI's Codex and Work Analytics dashboard at https://chatgpt.com/codex/cloud/settings/analytics#usage or run `/status` in the CLI. OpenAI provides no public personal-quota API for this site, so your countdown, notifications, and calendar export still run only in your browser. The public announcement feed remains available at `/api/feed`.