Codex Usage Limits: Check Reset Times, Credits and API Errors

Start by checking how Codex is authenticated. ChatGPT subscription usage and OpenAI API usage have different billing and limits. A subscription warning, an API rate error and an empty API credit balance can all interrupt work, but the same fix does not apply to all three.

Check your sign-in method and usage

In Codex CLI, use the commands below to inspect authentication and the active session. Do not paste API keys or stored credential files into a support ticket. In the desktop application, check the active account and workspace before comparing usage between devices.

codex login status

# Inside an active Codex CLI session:
/status

Subscription limits and reset times

OpenAI's pricing and usage guide describes five-hour estimates for Plus and Standard Business; Pro currently has no five-hour limit. Weekly limits may also apply. These are not fixed message counts for every task. Model choice and task size affect consumption, so consult the usage dashboard for your current reset times.

Eligible Plus and Pro accounts can buy additional credits. Workspace credit options depend on the organization's plan and administrator settings. Check what your account actually offers before assuming that changing plans or buying credits is necessary.

Before starting another large task, record the visible reset time and time zone. If you work across multiple accounts or organizations, record the workspace too. This avoids comparing a warning from one account with the allowance from another.

API rate limits are a separate problem

For API-authenticated work, check the HTTP status and error details. OpenAI's rate-limit guide distinguishes request and token limits. If a temporary rate error includes Retry-After, wait at least that duration and add a small random delay. Immediate retries can repeat the same failure.

An exhausted balance or enforced spend cap requires a billing action or the applicable billing reset, rather than rapid retries. The spend-limit guide explains the relevant error codes and controls. A ChatGPT subscription does not establish that an API organization has available budget.

A recovery checklist before changing billing

• Save your working files and inspect the diff. Preserve the current task instructions and remaining work.
• Record the warning text, time, account, workspace and model. Redact credentials before sharing diagnostic details.
• Subscription allowance reached: inspect the dashboard reset and the credit options available to that account.
• Temporary API rate limit: respect Retry-After when present and reduce request concurrency or batch size.
• API billing error: check the correct project and organization balance, budget and access permissions.
• No clear quota error: avoid changing billing as a diagnostic guess; collect the exact failure first.

Using an API key for local Codex work

The official authentication guide supports ChatGPT sign-in and API-key sign-in for local workflows. Codex cloud requires ChatGPT sign-in. API-key work uses OpenAI Platform billing and may have different feature availability and data policies.

For the official OpenAI API route, load a key into the environment and pass it through stdin as documented below. This is an authentication example, not a request to change your current session. Verify your budget and supported model before running paid work.

• Do not put a literal secret in a command you will paste into chat or commit to a repository.
• Keep the original configuration so you can restore the intended account and workspace.
• A third-party provider needs separate compatibility checks. It is not guaranteed to offer OpenAI models or every Codex tool and feature.

printenv OPENAI_API_KEY | codex login --with-api-key

Avoid losing progress during a limit incident

Write a short handoff note with the objective, modified files, checks already run and the next unresolved step. That note is useful whether you wait for a reset or resume with a different account. Review any partially applied changes before asking another session to continue.

When service resumes, try one small, reversible task first. Confirm that authentication, repository access and the expected tools work before assigning the full remaining job. Compare the result with the saved diff so the new session does not duplicate completed edits.

Sources and verification date

Reviewed on 15 October 2026 using official OpenAI documentation linked in each section. Limits and feature availability can change. Kintio is independent of OpenAI; this guide does not promise a way to reset or bypass an OpenAI allowance.

Frequently asked questions

Does every Codex account have the same weekly limit?

No. Use your account usage view and current plan documentation instead of applying one request count to everyone.

Does restarting Codex reset my allowance?

Do not rely on a restart to change an account-level allowance. Check the reset shown for the active account.

Is API billing included in my ChatGPT subscription?

API-key usage is billed through OpenAI Platform. Check that billing account separately before starting paid work.

Related guides

• Check OpenAI API billing and balance
• Diagnose an OpenAI API 429 response
• Evaluate alternative coding workflows

Last updated: 2026-10-15