How to set spending limits in OpenAI.

01The spending controls OpenAI gives you.

TakeawayTurn on Enforce a hard limit wherever traffic needs to stop, because an alert only warns.

02How to set an organisation spend limit.

1. Go to Organization limits in the API platform settings, at platform.openai.com/settings/organization/limits.
2. In the Spend section, select Edit spend limit.
3. Enter the Monthly spend limit.
4. Turn on Enforce a hard limit, so API requests fail once spend passes it.
5. Select Save.

You need permission to manage the organisation's settings. An organisation hard limit applies to API traffic across all projects, so it works as the ceiling for the whole account.

03How to set a project budget.

1. Go to Project settings and select Limits.
2. In the Spend section, select Edit spend limit.
3. Enter the Monthly spend limit.
4. Turn on Enforce a hard limit.
5. Select Save.

A project hard limit applies only to traffic billed to that project. The organisation limit and the project limit can both apply to the same request, and whichever is reached first stops it. A project limit covers everything billed to the project, so an agent or app that needs a budget of its own needs a project of its own.

TakeawaySet the organisation limit as the ceiling, then give each agent or app its own project with its own monthly limit.

04How to use spend alerts.

A spend alert sends a notification and leaves traffic running. OpenAI recommends alerts at thresholds that leave time to act before spend reaches a hard limit. Alerts stay active when a hard limit is added, so keep both: the alert gives warning and the hard limit holds.

05OpenAI usage limits by tier.

Separately from the limits you set, OpenAI approves a monthly usage limit for each organisation based on its usage tier. The tier moves up automatically as total credit purchases reach each threshold.

To see your tier, go to Settings, then Organization, then Limits. When the approved usage limit is the one you reach, request a higher one on the Organization limits page.

06What happens when a limit is reached.

A hard limit returns a 429 error on every affected request, and OpenAI warns that it can interrupt production traffic. Enforcement takes a moment to catch up, so the platform can process a small amount of extra usage while the limit state propagates, and recorded spend can end slightly above the limit. Traffic resumes once the limit is raised or removed and the change propagates, or when the limit resets with the next monthly cycle.

The error code tells you which limit was reached and what restores traffic.

• organization_spend_limit_exceeded or project_spend_limit_exceeded. Your own limit. Compare usage with the limit, then raise or remove it if the spend is expected.
• organization_usage_limit_exceeded. OpenAI's approved usage limit for your tier. Request a higher approved usage limit.
• credit_balance_exhausted. The prepaid balance is empty. Add credits.

TakeawayRead the error code first, because each 429 has a different fix.

07Usage limits in ChatGPT Enterprise and Edu.

ChatGPT workspaces are billed and capped separately from the API platform. Workspace owners and admins open the Admin Console, select the workspace and select Usage limits. There they set a default monthly limit for every user, a default for each group and an override for one person. A user override applies first, then the highest group default that applies, then the workspace default. Limits reset on the first day of each month in UTC, or at the start of each billing cycle if the workspace chooses that.

Members who approach their limit can request more from inside ChatGPT and Codex. Requests appear under Pending requests, admins receive an email, and an admin or owner approves the increase as temporary, ending with the current usage period, or permanent. In ChatGPT Enterprise, group managers with Manage spend controls switched on can approve requests for the groups they manage.

For the workspace as a whole, credit-based workspaces set an overage limit under Billing, by selecting Manage under Overage limit. OpenAI sets no overage limit by default, so it shows as No limit until an owner sets one. Token-based workspaces set a monthly budget in US dollars in the billing settings. Usage alerts in Billing notify the chosen recipients and leave usage running.

TakeawayIn ChatGPT Enterprise, set the workspace overage limit or monthly budget yourself, because a workspace starts without one.

08Where OpenAI's limits stop.

• OpenAI only. Spend on Anthropic, Google or a coding tool sits outside every OpenAI limit, so the company total is visible only when the invoices are added together.
• Two billing surfaces. API Platform limits and ChatGPT workspace limits are managed separately, and raising one leaves the other unchanged.
• Projects, with agents inside them. A project limit covers everything billed to the project, so agents sharing a project share its month.
• No request route on the API. OpenAI documents no way for a team to ask for more API spend. Whoever can manage the settings raises the limit, with no reason or expiry on record.
• A slightly late stop. Spend can pass the limit while enforcement catches up.

Our comparison of AI spending limits shows how OpenAI's controls sit beside Anthropic, Google, Azure, AWS, Cursor and GitHub Copilot, and how to set one budget above them all.

TakeawayTurn on OpenAI's hard limits as a backstop, and keep the budget for the whole company in one place above them.