Codex Usage Limits: Plans, Credits, Resets, and Token Usage

Codex usage is part of an account's agentic allowance, and the amount consumed by one task depends on its context, model, token mix, execution location, and duration. It should not be planned as one fixed number of coding messages.

Facts checked August 10, 2026. Limits and plan names change frequently. The official account page linked in each guide is the source of truth for the limits attached to your account.

Codex Usage Limits: The Short Answer

• Codex is included across eligible ChatGPT plans, but plan limits differ.
• Codex can share an agentic allowance with other supported agent features.
• Large repositories, long-running tasks, parallel agents, and higher-cost models use more capacity.
• The Codex Usage panel and limit banner show the options actually available to your account.

How to Check Codex Usage

Open Codex Settings → Usage in the Codex app or the Usage Dashboard on the web. Depending on the plan and workspace role, the panel can show limit utilization, remaining credits, recent usage, purchase options, auto top-up, or the next reset.

How Codex Credits Are Calculated

OpenAI's current flexible rate card maps Codex credits to input, cached input, and output tokens. Model choice matters because each model has its own credit rates. Cached input is generally cheaper than new input, while long output and higher-capability models can consume substantially more credits.

This token-based approach replaced the older habit of estimating every local or cloud task as one average message charge. Treat any per-task number as a planning example, not a guarantee.

What Happens When You Reach the Codex Limit?

An active turn may be allowed to finish subject to fair-use controls. Afterward, the account can offer one or more of these choices: wait for a reset, use a banked reset if eligible, add credits, enable auto top-up, or upgrade. The choices vary by plan and promotion.

Codex Usage vs OpenAI API Usage

Codex included in ChatGPT is not the same billing surface as direct OpenAI API calls. API usage appears in the OpenAI Platform Usage Dashboard by organization and project. Codex plan usage belongs to the Codex/ChatGPT account and its agentic allowance.

How to Reduce Codex Usage

• Give the agent a precise scope and acceptance criteria before it explores the repository.
• Keep project instructions concise and remove duplicated context.
• Use a less expensive capable model for routine changes.
• Avoid unnecessary parallel agents and unattended automations.
• Break ambiguous projects into independently verifiable tasks.

Codex Limits FAQ

Does every Codex message use the same amount?

No. Context, model, cached input, output, task complexity, and execution mode all affect usage.

Where is the reset time?

Use the Usage panel or limit banner. Do not infer it from another user's plan.

Can Plus and Pro users buy more Codex usage?

Eligible accounts can be offered flexible credits. Availability and options are shown in the Usage area.

Official references: Codex with a ChatGPT plan · Codex rate card.