Codex pricing and limits vary by plan, workspace, model, task size, and rollout. AIPlanWatch records the official source and keeps uncertain quotas explicitly unverified.
What the official sources say
OpenAI's pricing page lists ChatGPT plan prices, while the Codex rate card explains token-based credit rates. The Codex usage guide says limits depend on plan and task complexity.
Why the numbers are not a single quota
A Codex task can consume different credits depending on input tokens, cached input, output tokens, model, reasoning mode, and whether agents run. This page does not turn a rate card into a fixed message allowance.
How to verify a change
Check the linked OpenAI source, record its checked date, and distinguish an official plan price from an observed usage report. Reset notices can have eligibility and expiration rules, so publish them only after the exact date and scope are confirmed.
Editorial policy
AIPlanWatch publishes a fact only after an editor confirms the source, date, scope, and plan. Draft records remain private until those checks pass.
Frequently asked questions
Are Codex usage limits fixed?
No. Official guidance says usage depends on plan, task complexity, model, and current workspace eligibility.
Sources and verification
Last checked Aug 13, 2026, 9:17 AM UTC.