Codex Usage Calculator: Will Your Quota Last Until Reset?

Enter your weekly quota remaining, time until reset, and the change between two recent usage readings. The calculator gives you a safe daily budget, your observed burn rate, and whether your current pace reaches 0% before the weekly reset.

No hidden cap guess

Uses your account's displayed percentage, not an invented token ceiling.

Two readings

Measure the meter twice to turn recent real usage into a pace forecast.

Local only

The calculation runs in your browser. Your quota values are not uploaded.

Weekly quota runway

All calculations happen in your browser. Nothing is uploaded or saved.

Enter two usage readings

Check your weekly meter, work for a while, then check it again. Enter the percentage point drop and the time between those readings.

Weekly quota remaining now

Use the weekly remaining percentage shown by /status or Settings > Usage.

Time to weekly reset

days

Plus

hours

Meter drop

points

Example: 72% to 64% remaining is an 8 point drop.Time between readings

hours

At this pace, you run out before the weekly reset.

Projected exhaustion in 1d 21h, about 1d 3h before reset.

Safe daily budget

20.0%

of weekly meter per 24h

Current pace

32.0%

of weekly meter per 24h

Runway

Time until weekly reset: 3d
Time to 0% at current pace: 1d 21h
Observed burn rate: 1.33 points/hour

Pace change needed

Reduce your observed burn rate by about 37.5% to make the current weekly allowance last until reset.

Diagnose fast usage drain

This is a pace forecast, not a prediction of OpenAI's hidden metering formula. Your real burn rate can change when the model, reasoning level, context size, tool loops, or workload changes.

How to get the four numbers

1
Check /status in Codex CLI, or open ChatGPT Settings > Usage. Note your weekly percentage remaining and the reset time.
2
Work normally for a useful sample. An hour or more is usually more informative than a few minutes because a very short interval can exaggerate noise in the displayed meter.
3
Check the weekly meter again. If it moved from 72% remaining to 64%, your meter drop is 8 percentage points.
4
Enter the 8 point drop and the hours between readings. The calculator compares that observed pace with the pace your remaining quota can sustain until reset.

What the calculation actually tells you

The useful comparison is simple. Your safe pace is the percentage remaining divided by the hours until reset. Your observed pace is the percentage point drop divided by the hours between readings. If observed pace is higher, the current workload exhausts the meter before the reset unless your usage slows.

Why percentages are enough: this tool is not estimating tokens. It is answering a narrower question: whether the pace at which your own displayed weekly meter is moving can be sustained for the time left in the current window.

When to recalculate

Recalculate after a meaningful workload change. Our first-party session analysis found that Codex input processing can vary dramatically even when human prompt count looks similar, because context carryover, tool loops, compaction, and model behavior can dominate the session. A pace measured during one workload is not a permanent property of your account.

See the 299-session study•Diagnose fast usage drain•Understand the weekly limit

Want the meter tracked automatically?

The calculator is for a quick browser check. Codex Usage can track your current quota, reset countdowns, session pace, and runway locally so you do not have to keep taking manual readings.

Install the free local tracker

Codex usage calculator FAQ

How does the Codex usage calculator work?

It uses four values you can observe directly: your weekly percentage remaining, time until reset, the percentage point drop between two meter readings, and the hours between those readings. From those values it calculates your current pace, safe daily budget, and projected time to zero.

Does this calculator know my exact Codex token limit?

No. OpenAI does not publish an exact weekly token ceiling for subscription plans. The calculator deliberately avoids guessing one. It works from the percentage meter shown for your own account, so the forecast is about pace against your displayed allowance.

Why can the forecast change after I calculate it?

Codex usage is workload dependent. Changing models, reasoning level, context size, tool use, or task complexity can change how quickly the displayed meter moves. Recalculate from two recent readings when your workload changes materially.