Search topic
codex usage limits weekly 5 hour
Practical outcome
Where to check Codex's 5-hour and weekly limits—and what local token history cannot tell you.
Quick answer
/status, in ChatGPT Settings, or on OpenAI's usage dashboard. A local token tracker can explain your past activity, but it cannot calculate how much ChatGPT allowance remains. ⏳short usage window
separate allowance
usage per task
Codex access on ChatGPT plans is not a simple public token bucket. OpenAI describes plan usage in terms of the work Codex performs, and that consumption changes with the selected model, context size, reasoning effort, task complexity, tools, and whether work runs locally or in the cloud. That is why two sessions with the same number of prompts can consume different portions of the allowance.
The authoritative place to see the allowance is OpenAI's own product UI. Open Codex and run /status, or check Settings and the usage dashboard. OpenAI can change limits, plan treatment, and reset options, so this guide links to the current documentation instead of inventing a universal token threshold. The details below were reviewed on August 28, 2026.
Do the work
model, context, tools
Check allowance
/status or usage dashboard
Review history
local daily/model totals
What are the Codex Plus limits?
The two plan counters apply at the same time. Reaching either one can prevent new work until capacity becomes available again. Do not rely on a hand-built calendar that assumes a fixed weekly reset day: use the reset information shown in the account. OpenAI also notes that an active turn may continue subject to fair-use protections, so “the current task always aborts at the cap” is not a safe rule.
The allowance may be shared
OpenAI's current guidance says Codex can share an allowance with other agent experiences such as Work, Excel, and Workspace Agents. If the counter moves faster than your terminal history suggests, check activity in every product attached to the same ChatGPT plan.How to check Codex usage now
1. Ask OpenAI for the live counter
Run /status inside Codex, then compare it with the usage surface in ChatGPT Settings. Those product counters determine whether more plan work is available.
2. Read the displayed reset
Record the reset date or time that OpenAI shows. Avoid assuming a calendar boundary, and re-check after buying or banking any reset option because the displayed date may change.
3. Diagnose historical burn locally
Run npx whoburnedmore --local to inspect supported local logs without signing in or uploading aggregates. Use the daily and model totals to find large sessions—not to predict the subscription counter.
4. Separate allowance from cost
A ChatGPT subscription limit is not the same thing as an API invoice. Any model-price calculation from local tokens is an API-equivalent estimate, not your remaining Plus allowance or the amount billed by ChatGPT.
Why Codex usage varies by task
A short question over one file and a repository-wide migration are different workloads. Long context, repeated tool calls, high reasoning settings, large command output, retries, and cloud execution can all increase consumption. The useful planning unit is therefore a completed task with a known scope, not an assumed number of prompts or a made-up token ceiling.
Compare similar tasks over several days. If repository-wide requests repeatedly coincide with large local totals, narrow the file set, start a fresh session when stale history dominates, or split the work into independently verifiable steps. That can improve predictability, but it still does not convert raw tokens into a guaranteed percentage of a ChatGPT limit.
What happens at the cap?
The product tells you when new plan-backed work is unavailable and when capacity should return. Depending on the account and the options OpenAI currently offers, the practical choices may include waiting for the displayed reset, using an available credit or reset path, changing plans, or moving separately billed work to the API. Check the account UI before taking action; plan names do not imply one universal quota structure for every customer.
Is Codex limited for ChatGPT Plus?
Yes. Plus includes Codex access with usage limits. The live product surfaces show the account's current short-window and weekly state.
Can whoburnedmore show my Codex weekly limit?
It can summarize historical tokens present in supported local Codex logs. It cannot read or reverse-engineer the private ChatGPT entitlement formula, include cloud-only work that is absent from those logs, or guarantee when OpenAI will allow the next request.
Where is the official Codex usage page?
Start with OpenAI's current Codex plan guidance, then use /status, ChatGPT Settings, or the linked usage dashboard for your account. For a historical token view, see the Codex CLI usage guide.
Related guides
How to Check Codex CLI Usage and Token Cost
Codex CLI's /status only shows the current window — here's how to see your full history and cost.
Claude Code Rate Limits, Explained
The weekly and 5-hour caps demystified — plus how to see your real utilization.
OpenAI Codex CLI Cost: Estimate and Track It
The token math behind Codex CLI pricing — plus how to track what you really spend.