Direct answer
OpenAI’s Help Center says ChatGPT agent is no longer available. The same page still lists legacy monthly limits of 40 messages for Plus, 400 for Pro, and 40 for Business and Enterprise. Because the feature is marked unavailable, these figures are historical information—not a reliable statement of current access or an allowance you can use to choose a plan today. (OpenAI Help Center)
For longer, multi-step work in ChatGPT, OpenAI directs readers to ChatGPT Work. Work does not have a fixed allowance stated as a number of messages or tasks: the amount of work available depends on the model, task, and settings. Check Settings → Usage for your remaining allowance and reset times. (OpenAI Help Center · Work and Codex usage)
| What you’re using | What the documentation says | Where to check |
|---|---|---|
| ChatGPT Agent mode | The Help Center says it is no longer available. The same page’s 40/400/40 monthly figures are legacy limits, not confirmation of current access. | Check current product availability in your ChatGPT account. |
| ChatGPT Work | Allowance is not a fixed number of messages or tasks; usage depends on the model, task, and settings. | Settings → Usage for remaining allowance and reset times. |
| OpenAI API | API rate limits are separate from ChatGPT subscription allowances. They vary by model and apply at organization and project level. | Check your applicable limits in the developer console. |
How to interpret the old Agent mode caps
The figures—Plus: 40 messages/month; Pro: 400 messages/month; Business and Enterprise: 40 messages/month—appear on the Help Center page that also says ChatGPT agent is no longer available. Treat them as legacy limits, not evidence that Agent mode is currently accessible on those plans or that upgrading restores access. The page does not establish a current Agent mode allowance or reset rule. (OpenAI Help Center)
What to expect from ChatGPT Work
OpenAI describes Work as the ChatGPT option for longer, multi-step tasks and finished deliverables. Its allowance is not a one-task-equals-one-message quota: how much work it covers depends on the model, task, and settings. Check Settings → Usage for your account’s remaining allowance and reset times rather than trying to convert the old Agent mode caps into a number of Work tasks. (Work and Codex usage)
ChatGPT limits are not API limits
A ChatGPT plan and an API project use separate limit systems. OpenAI documents an Agents API in beta for developers building applications with an OpenAI-managed Codex agent that can work across multiple steps and continue in the same session. This is a distinct developer offering; the documentation does not establish feature parity with the former ChatGPT Agent mode. (Agents API beta FAQ)
For API requests, rate limits vary by model and are defined at the organization and project level. Check the applicable limits in the developer console, and review current API pricing separately before choosing an API workflow. Do not apply ChatGPT subscription message figures to API usage or treat API rate limits as Agent mode or Work allowances. (OpenAI API rate limits)
Which route fits? Use ChatGPT Work if you want to do multi-step tasks within ChatGPT; check your account’s usage display for the allowance and reset times that apply. If you are building an application, investigate the documented Agents API beta and its Codex-agent scope, then check your project’s API limits and pricing. Do not assume the API option reproduces the former ChatGPT Agent mode experience.
What to check before choosing
- If you mean the old ChatGPT Agent mode: don’t budget around the 40/400/40 figures. The Help Center labels Agent unavailable while retaining those legacy limits.
- If you mean multi-step work in ChatGPT: open Settings → Usage to see your remaining allowance and reset times. Usage depends on the model, task, and settings.
- If you are building with the API: review the Agents API beta documentation, check organization- and project-level limits in the developer console, and consult API pricing for your intended use.
Sources
ChatGPT agent — OpenAI Help Center
Managing usage with GPT-6 Astra in Work and Codex — OpenAI Help Center