Codex reset statistics: how public signals behave
Numbers drawn from the 37 public Codex reset signals this radar has logged since April 28, 2026. They describe public signal patterns, not your account — your Codex UI remains the source of truth.
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.
37
Logged since April 28, 2026 — 30 high confidence, 7 medium confidence.
4.9 days
Between consecutive logged signals. Includes 6 same-day repeat signals — each counted as an event, 0-day gaps skipped in the average.
2.4 days
Across 15 signals in the last 30 days. A short average does not predict the next reset — signals cluster around product events.
Which weekdays signals land on
Counts by weekday of the signal date, as logged (date-only, no time-of-day data exists for these signals).
What typically triggers a public reset signal
Grouped by public signal type, not by official reasons — OpenAI does not publish reset reasons. How types and confidence labels work: Methodology.
View the full signal history → Get the next reset alert →
Sources & citations
The reset windows and status signals shown above are derived from publicly available sources. We do not invent figures and we do not have private access to OpenAI's systems. The primary references we lean on:
OpenAI official documentation
- Introducing Codex — OpenAI's product page for the Codex coding agent. Used as the canonical definition of what "Codex" means in our copy.
"Codex is powered by codex-1, a version of OpenAI o3 optimized for software engineering."
- Using Codex with your ChatGPT plan — OpenAI Help Center article covering how Codex usage limits work when signed in with a ChatGPT account, including the explicit note that support does not reset usage limits.
"OpenAI Support does not reset ChatGPT or Codex usage limits. If you reach a limit, wait until it resets or use another available option shown in your account."
- ChatGPT agent — usage & limits — Help Center article explaining how tier-based monthly message limits and reset windows are defined for ChatGPT-side features that share the Codex quota surface.
"Agent mode includes the following monthly message limits … subject to reasonable rate limits … in order to ensure the product works well for all users."
- ChatGPT — Release Notes — the running public changelog where OpenAI announces limit changes; we treat every "weekly" or "monthly" wording edit here as a signal that the reset window definition may have moved.
OpenAI status page
- OpenAI Status — root — the live status feed for ChatGPT, the API, and Codex. We monitor this feed for elevated-error incidents that historically line up with usage-side anomalies.
- OpenAI Status — History — the public archive of every past incident, useful for backfilling the history page.
- Incident example — "Elevated errors across ChatGPT and Codex", 2026-09-03 — a recent public incident cited as a representative anchor for what a Codex-affecting event looks like in the status feed.
"The issue causing elevated errors across ChatGPT and Codex has been resolved. Some Codex remote control users may need to pair their mobile device again following this incident."
OpenAI terms & policies
Used as the legal/definitional anchor when we describe how Codex is offered. We link, not quote, here.
- Terms of Use — openai.com/policies/terms-of-use
- Privacy Policy (international) — openai.com/policies/row-privacy-policy
Public community signals
Beyond official sources, our methodology cross-checks against publicly visible community chatter on X (formerly Twitter) and Reddit. We use these platforms only as a sentiment and corroboration layer — they are never treated as ground truth and we do not link to individual posts in this section. If a public thread has been material to a specific published estimate, that thread is cited inline on the relevant page.