Codex Reset Radar
🌐 English

Codex "exceeded retry limit, last status: 429": what it means

The Codex error "exceeded retry limit, last status: 429 Too Many Requests" explained with real GitHub issues: when it is a platform incident, when it is your own short rate window, and what to check in order.

Short answer: This message means Codex received an HTTP 429 (Too Many Requests) and stopped. Despite the wording, the client often has not retried at all — and the cause is either a platform-side incident or a short-term rate window on your account, not necessarily your weekly quota.

This is an unofficial tracker. It is not affiliated with OpenAI. Codex is a product/trademark of OpenAI. Probability scores are public-signal heuristics, not authoritative announcements or certainty.

What the message actually means

The full error usually reads exceeded retry limit, last status: 429 Too Many Requests, request id: … — exactly as filed in openai/codex issues #22122 and #26048. A source-level trace in issue #30471 shows why the wording misleads: for several request paths the client's retry_429 flag is hard-coded to false, so the first 429 is mapped straight to a RetryLimitReachedError and displayed as "exceeded retry limit" — without a single retry having happened. Read it as "request rejected with 429", not "I retried too many times".

Platform incident, or your own account

Both happen, and they need different responses. Platform side: on 2026-06-03 a burst of these errors was confirmed by OpenAI staff as an active incident (see #26038, closed as a duplicate of the incident thread #26034; #26040 and #26048 were filed in the same window). Account side: users report 429 loops while usage panels still show plenty left — #54010 documents repeated 429s with 69% allowance remaining and only 3 active sessions, and #23304 a Pro plan hitting 429 loops mid-normal-work. A 429 with a usage-limit response body is a different error (your quota), which is why you should check which one you are actually hitting.

What to check, in order

One: start a fresh Codex session and retry once — capacity blips clear in minutes. Two: run /status inside a Codex session and read both limit bars — if the 5-hour or weekly bar is genuinely empty, this is a quota event, not a mystery 429 (see our how to check Codex usage guide). Three: check status.openai.com and the reset radar below — if other users are reporting the same wall of 429s at the same timestamp, it is platform-side. Four: retry from a different network — VPN ranges and shared egress IPs can carry an extra throttle that has nothing to do with your plan. Five: if it persists with quota remaining, file an issue with the request id from the error.

What not to do

Do not hammer retry in a tight loop — every retry is another request against the same short window, which can deepen the throttle (maintainers make the same point in #30471). Do not paste your request id, email, or workspace name into public screenshots without trimming them. Do not treat one Reddit post as proof of a global incident. And no tracker — including this one — can see your personal quota; never enter Codex credentials into a third-party "429 fixer" tool.

Is it a platform-wide event, or just your account?

The same error text covers two very different situations. Before you spend an hour debugging your setup, check whether the timing matches a public reset or incident window — that is exactly what an unofficial public-signal radar can help with, and everything else on this page is between you and your own Codex UI.

Latest public signal (radar snapshot)

Latest public signal: Tibo on X: banked reset activated — lands in everyone's accounts soon (Nov 12) (2026-11-12) · Public signals logged: 64

Radar monitor

Last radar sweep: 2026-11-22 15:15 UTC. The radar watches public posts only — it never sees individual accounts, rate windows, or 429 responses.

Public signals are context, not account-level proof. Your Codex UI and status.openai.com remain the sources of truth. See the live radar →

Two signatures, two conclusions

Everyone at once, request ids in sequence, status page degraded — treat it as platform-side: pause heavy work, do not burn a banked reset, and wait for resolution. This matches the 2026-06-03 window where OpenAI staff confirmed an active incident (#26038, #26034).

Only you, quota bars showing room, same network — treat it as account/network-side: try a fresh session, then a different network, and read your /status bars before filing anything. This matches reports like #54010 (69% remaining) and #23304 (Pro, mid-normal-work).

Sources: real reports behind this error

Every claim on this page traces to a public report. Error text verbatim: #22122, #23304, #26048. Source-level explanation of the misleading wording: #30471. Confirmed platform incident (2026-06-03): #26038 (closed as duplicate of the incident thread), with #26040 in the same window. 429 with allowance remaining: #54010. Early reports: #11434, #11083. These are community/public reports, not OpenAI documentation.

Want the next strong Codex reset signal?

Get an email when the radar logs a stronger public reset signal.

Get reset alerts →

Codex "exceeded retry limit, last status: 429": what it means FAQ

Is "exceeded retry limit, last status: 429" the same as hitting my usage limit?

No. A 429 is an HTTP-level rejection; your usage limit is a quota state. Issue #54010 shows repeated 429s with 69% allowance remaining. That said, a 429 whose body carries the usage-limit shape is surfaced as a usage-limit error instead — so check your /status bars before assuming either one.

Does this error mean Codex is down?

Not necessarily. On 2026-06-03 OpenAI confirmed an active 429 incident (#26038), but the same message also appears for account-level short-window throttling. Cross-check status.openai.com and public reports before concluding it is an outage.

When does the 429 clear?

If it is a short-term rate window, it clears when that window rolls forward — minutes to a few hours, not the weekly reset. If it is a platform incident, it clears when OpenAI resolves it; the 2026-06-03 burst was fixed within the day.

Should I keep retrying until it works?

No. Retry once after a pause in a fresh session. Rapid retries add load to the same rate window and can make the throttle stickier. If /status shows quota remaining and retries still fail, collect the request id and report it.

Can Codex Reset Radar tell me why I got a 429?

No. This tracker only logs public reset signals for platform-level context. It cannot see your account, your rate windows, or your request ids, and it will never ask for credentials.

Related Codex reset guides