When throttled, start in the Usage Dashboard: read the five-hour and weekly indicators plus the exact reset times. Use CLI /status for a quick remaining-limit check.
OpenAI does not publish a universal per-message replenishment formula, so this site does not infer one from activity five hours or seven days ago.
The decision tree is simple: short allowance low → schedule after the displayed time or switch to lighter work; weekly allowance low → wait, evaluate eligible credits or instant reset, or compare plans; both healthy → investigate model, network and task conditions.
The compliance line remains firm: no multi-accounting, automated bypasses or shared-account tactics.
Which limit are you hitting?DECISION TREE
Five-hour allowance is low
Schedule around the Usage Dashboard time and switch to lighter work.
Read the five-hour guide →Weekly allowance is low
Stop guessing the formula; use the account date, then weigh waiting or a plan change.
Read the weekly guide →A reset credit or Instant Reset appears
Redemption changes allowances and the reset date; read the confirmation first.
Read reset-credit rules →You do not know which limit binds
Check remaining limits with /status, then copy the Usage Dashboard time.
Check personal usage →Limits repeatedly block delivery
Measure a full week of real blockers before comparing Plus, Pro and Business.
Compare plans →FAQFAQ
Can both limits be exhausted at once?
The Usage Dashboard can show both allowances. Read the two indicators directly rather than assuming which one must bind first.
Does switching models help while throttled?
Very much so: heavy models burn allowance far faster than lighter ones. Routing routine tasks to lighter models and saving the heavy allowance for hard problems is the most effective short-term relief.
Is the 5-hour limit still in force?
It was temporarily removed starting July 13, 2026; policies have changed before and may again. Trust official announcements and your account's actual behavior — this site's event pages record every policy phase.