Codex vs. Claude Code usage limits: what's actually different
Updated 2026-09-22
“ChatGPT Plus codex limits vs Claude Code” is a common search because the two products handle usage limits differently enough to be confusing. This is a structural comparison, not a numbers comparison — for the actual figures, check each vendor's own current pricing page.
Global-reset totals via whenreset.app as of the latest build: Codex has 56 recorded global resets (last on 2026-10-26), plus 8 banked credits not counted in that cadence; Claude has 14 recorded global resets (last on 2026-10-21), plus 1 banked credit not counted in that cadence. These numbers are pulled fresh at build time, not typed into this guide by hand.
Both use rolling limits, tied to different things
Codex usage is tied to your ChatGPT plan tier (Plus, Pro, Business, and so on) and rolls on a 5-hour and weekly cycle. Claude Code draws from the same Pro/Max usage pool as claude.ai itself — see the Claude Code usage-limit guide for more on that. Neither vendor publishes the underlying numbers behind these windows in a way this site tracks, so this page intentionally doesn't restate figures that live on the vendor's own site and can change without this page being updated in sync.
Where to check your real numbers
For Codex, run /status in the CLI. For Claude Code, run /usage, or check Settings > Usage on claude.ai. Both of these reflect your actual account state; nothing on this site can substitute for them, since this site only tracks publicly announced global resets, not individual usage.
How resets are announced differently
Codex's global resets tend to be announced individually by OpenAI Codex lead @thsottiaux. Claude's are split between Anthropic's official @ClaudeDevs account and members of the Claude Code team, and — based on this site's own recorded totals — have come less often so far. See the live “Total recorded” and “Typical gap” figures on the Codex and Claude pages for the current comparison, since which one resets more often can shift over time.
Picking a tool based on limits alone probably isn't the right call
Usage-limit behavior is one input among several — model quality for your specific task, editor/CLI integration, and pricing all matter too. If you're deciding based on running out of usage specifically, the what to do when you hit the limit guide covers practical options that don't require switching products at all.
Frequently asked questions
Which one resets more often, Codex or Claude Code?
Check the live “Typical gap” stat on the Codex and Claude pages rather than trusting a fixed answer here — the relative cadence between the two has shifted before and can shift again.
Do ChatGPT Plus and Claude Pro include the same amount of usage?
There's no shared unit to compare them on directly from this site's data — check each vendor's own current plan page for what's included. This site only tracks reset events, not plan-to-plan limit comparisons.
If I hit the limit on one, does switching to the other help?
It can buy you time short-term, but it doesn't change how either product's limits work. See what to do when you hit the limit for options that don't depend on switching products.
Back to Codex reset history · Back to Claude reset history · Subscribe: all-products feed