Codex usage limits: how to check, when they reset, and what to do

Understand Codex five-hour and weekly usage limits, check remaining usage and reset time, and compare waiting, plan options, paid resets, and API-key billing.

Codex usage limits: how they work, how to check, and what to do when you hit them

Codex usage limits are not a single fixed number of prompts. Your available usage depends on your ChatGPT plan, the model, the task, and whether a five-hour or weekly limit applies. Here is how to check the limit on your account, understand the reset time Codex shows, and choose what to do next.

People searching for a Codex weekly limit or a Codex limit reset often mean different things: a short-window cap, a weekly allowance, or the account’s next displayed reset. The steps below help you tell them apart.

Official information checked 2026-09-23. Plans and usage details can change. The Usage page and the notices shown in your Codex account are the source for your current allowance and reset time.

The short version

• Run /status in the Codex CLI to see the active session’s remaining usage information.
• For the account’s current allowance and reset time, open Settings → Usage in ChatGPT. Official guidance also directs Codex users to the account usage view.
• A five-hour allowance and a weekly limit are different windows. A weekly limit may apply in addition to the shorter window.
• There is no universal reset hour that applies to everyone. Use the reset time shown for your account.
• Your ChatGPT plan’s included Codex usage and API-key billing are separate. An API key uses pay-per-use API billing.

What “Codex limit” means

Codex can enforce more than one usage window. Current plan information describes a shorter, five-hour usage window and says weekly limits may also apply. The plan page’s message counts are estimates, not a promise of a fixed number of tasks: model choice, task size, context, and other usage details can change how quickly a person reaches a limit. Cloud and local Codex work draw from the included allowance described for the plan.

That is why “How many Codex prompts do I get?” has no one number that fits every account. The plan page publishes plan-specific estimates, and the actual remaining amount is account-specific. Check the current Codex plan and usage page and official usage guidance for current plan details.

How to check remaining Codex usage

In the Codex CLI

While Codex is running, enter:

/status

This displays usage and reset information available to the current CLI session. If you are not signed in, or use a different authentication method, what appears may differ.

In ChatGPT account settings

Open ChatGPT Settings → Usage to see the allowance and reset information for your account. This is the best place to answer “when can I use Codex again?” because it shows your account’s own current state. If Codex displays a limit notice, read the full notice; it may identify whether you reached a short-window limit or a weekly limit and show available next steps.

If you use an API key instead of signing in through a ChatGPT plan, check the billing and usage information for the API account that issued the key. ChatGPT plan usage and API usage are billed separately.

When does a Codex usage limit reset?

The reset time depends on which limit you hit and your account’s usage window. Do not assume every account resets at midnight, on Monday, or at the same time zone. official guidance says the Usage view shows the current reset time; follow that timestamp rather than a schedule copied from a forum post.

For eligible personal Plus and Pro accounts, Official materials also describe a paid option to reset a weekly Work and Codex limit immediately. Availability can vary. This moves the normal weekly allowance forward; it does not add a second weekly allowance. The new weekly period starts with the first Codex or Work request after the reset and the next reset is seven days later. A five-hour-only limit notice does not by itself trigger the in-app weekly reset option. Check the official paid weekly reset article to see whether the option applies to your account.

What to do when you reach a limit

1. Wait for the displayed reset

If you can pause, stop sending requests and resume when the time shown in Usage arrives. Starting more tasks or repeatedly retrying may consume more of the available allowance without solving the limit.

2. Review plan options

If your work regularly exceeds the included allowance, compare the current plan options and their published usage estimates. An upgrade may provide a higher allowance, but the precise amount depends on the plan, model, account, and current official materials policy. Read the current account offer before changing plans.

3. Use an offered paid reset or credits, if available

Some eligible accounts may see an option to reset a weekly limit or add credits. These options are not displayed for every account or every kind of limit. Read the price and terms on the confirmation screen before purchasing. A paid weekly reset advances the ordinary allowance; it is not an additional weekly quota.

4. Use a pay-per-use API key

An API key is a separate billing route from a ChatGPT subscription. Codex documents API-key authentication and custom model providers. You can use the API or, where the model and required Codex workflow are supported, an API-compatible provider such as BazaarLink. This can be useful if you prefer usage-based billing, but it is not the same as extending a ChatGPT plan: some ChatGPT workspace and cloud features may not be available with API-key authentication. Read the official Codex authentication and custom-provider configuration guidance, and check the provider’s available models and rates first.

Example: configure a BazaarLink provider

The following example uses Codex’s custom provider configuration and BazaarLink’s documented API-compatible endpoint. It uses the Responses API protocol:

model = "gpt-5.3-codex"
model_provider = "bazaarlink"

[model_providers.bazaarlink]
name = "BazaarLink"
base_url = "https://api.bazaarlink.ai/v1"
env_key = "BAZAARLINK_API_KEY"
wire_api = "responses"

Set BAZAARLINK_API_KEY in your shell environment; keep the value out of the TOML file, source code, and chat logs. The model gpt-5.3-codex appeared in BazaarLink’s live model catalogue on 2026-09-23 at US$1.75 per million input tokens and US$14 per million output tokens. Catalogue models and prices can change. The amount billed depends on actual input and output tokens and the model’s current rate. A catalogue listing does not guarantee that every Codex feature or workflow works with that provider; verify the exact model and workflow before relying on it.

For a broader comparison of subscription and usage-based billing, see our subscription versus usage API guide. For a Codex-specific price discussion in Traditional Chinese, see Codex 費用怎麼算.

Official sources

• Codex plan information and usage estimates
• Managing usage in Work and Codex
• Using Codex with a ChatGPT plan
• Paid weekly Work and Codex limit resets
• Codex authentication
• Codex custom providers

Last checked: 2026-09-23.

TWD billing · Taiwan invoices · leading AI models · OpenAI-compatible API