Overview

Token-based billing is a billing model for eligible ChatGPT Enterprise plans. Usage across Chat, Work, and Codex is metered in tokens or other units specified in the rate card and charged in US dollars (USD) at the rates in your agreement. Usage charges apply alongside any contracted seat fees.

This model changes how usage is measured, reported, and billed - feature access continues to follow your plan and workspace permissions.

Availability

Token-based billing applies to ChatGPT Enterprise workspaces with an eligible agreement, including ChatGPT for Financial Services. It is not currently offered for ChatGPT Edu or ChatGPT for Healthcare.

Your account team can arrange token-based billing through an eligible new or replacement Enterprise agreement. Your agreement determines which billing model applies. Some ChatGPT Enterprise workspaces continue to use credit-based billing. If you are unsure which model your workspace uses, contact your OpenAI account team.

How token-based billing differs from credit-based billing

Credit-based Enterprise agreements

Traditional credit-based Enterprise agreements use a workspace credit balance for metered usage. Depending on the agreement, credits can come from recurring seat allocations or purchased credit pools. As credits are used, the balance decreases. An overage limit can control additional usage after the available balance is exhausted.

Metering can differ by product: Chat usage can be charged per message, while Work and Codex use token-based metering. Included usage and applicable credit rates depend on the agreement.

For more information about credit-based plans, see: Flexible pricing for ChatGPT Enterprise and Edu, and see the ChatGPT Rate Card (Credit based pricing).

Token-based Enterprise agreements

Usage across Chat, Work, and Codex is metered in tokens or other units specified in the rate card, and the cost is shown in dollars. In contrast with credit-based billing agreements, which draw down from a workspace credit balance that was purchased upfront, usage in token-based billing contributes to a monthly invoiced amount during the billing period.

There may be additional variations in your plan, compared to credit based billing. Note that a previous credit total is not a direct estimate of the same activity's dollar cost. Actual costs may vary based on usage. For model and usage rates, see: ChatGPT Rate Card (Enterprise token-based pricing). Your agreement determines the rates and discounts that apply to your workspace.

Seat fees, commitments, and invoices

Token-based usage charges are separate from any seat fees in your contract. Where seat fees apply, the number of assigned seats at the start of each billing month sets your baseline seat charge, even if assigned users haven’t sent a message or run a task. Any additional seats you add are prorated—charged proportionally for the remaining part of the month.

Removing seats lowers your assigned seat count immediately, but reduces your billable baseline starting the next billing month. Seat fees for the previous month are billed after each monthly billing period ends. Your account team can confirm your rates and any contract-specific terms.

An eligible agreement can include a spending commitment across ChatGPT Enterprise and the API. A commitment is an agreement to spend a specified amount over the contract term. It is not a prepaid workspace credit balance. Usage is invoiced as it occurs over the term, and any shortfall against the commitment is billed according to your agreement.

ChatGPT usage and any contracted seat fees are invoiced monthly in arrears, meaning after the billing period ends. API usage is invoiced separately. Monthly invoicing does not make the agreement month-to-month; the contract establishes its term and payment obligations.

For billing periods and invoice guidance, see: Understanding invoice dates.

Spend controls and reporting

Admins can manage dollar-based usage with a monthly workspace budget and user and group limits. The workspace budget controls spending across the workspace, while user and group limits govern individual allowances. Spend alerts help admins monitor usage; an alert does not increase a budget or authorize additional spending.

For setup and ongoing management, see: Manage usage limits and overages in ChatGPT Enterprise and Edu. Follow the guidance for token-based Enterprise agreements.

Usage reporting continues to show token information, with dollar amounts replacing credit amounts for current spend. If your workspace previously used credits, earlier credit history remains available separately.

ChatGPT reporting shows ChatGPT spend. API spend remains in the API platform, even when both products count toward the same contractual commitment. A ChatGPT spending report therefore does not show your total progress against a commitment across both products.

For workspace administration guidance, see: ChatGPT Enterprise admin quickstart.