SHORT WINDOW
5-hour / session limit
This is the shorter operating window shown in Codex. It can recover before the weekly allowance does, so a temporary stop does not necessarily mean the weekly quota is exhausted.
WORKING REFERENCE
Codex Usage Limits
A stable mental model for the limits you see—kept separate from values that can change.
Last updated: September 15, 2026LONG WINDOW
Weekly limit
The weekly allowance is the longer runway. A healthy short window cannot override an exhausted weekly allowance; check both whenever work is blocked.
VARIABLE COST
Why tasks consume differently
Codex usage can vary with the selected model, task complexity, context size, and other service conditions. Treat the indicators in your account as the source of truth.
SEPARATE SYSTEM
Subscription vs. API
Codex access included with a ChatGPT subscription is separate from usage billed through the OpenAI API. Reaching one limit does not describe the balance or billing state of the other.
CHECK / STATUS
Where to check usage
Run /status inside Codex or use the current usage view provided by the product. Copying that output into this site’s Usage Tracker keeps the parsing on your device.
Treat the percentages and reset times displayed in your own account as authoritative for your subscription. This community tool does not query or inspect your account.
FAQ / LIMITS
Common questions
Where can I check my Codex usage?
Open Codex and run /status, or check the usage information shown by the product. Availability and wording can change over time.
Are the five-hour and weekly limits fixed token amounts?
Do not assume they are simple token counters. Consumption can vary by model, task, context, and current product rules.
Is subscription usage the same as API usage?
No. Subscription access and API billing are separate usage systems with separate controls and limits.
NEXT / PATH