API reference
Billing
How your balance and charges work. Current prices are on the Pricing page.
Prepaid balance
Your account has a balance in USD. Top up with a card on the Billing page. There's no subscription, and unused balance doesn't expire.
How a chat call is charged
Each reply is charged for the exact input and output tokens the model reports, the same numbers you see in the response's usage. A call is refused with 402 only if your balance couldn't cover a reply of max_tokens length.
Failed calls
Failed calls and refused requests are refunded in full.
Volume discounts
High-volume customers get discounted rates. Email us with your expected traffic.
Invoices
Every top-up comes with a Stripe invoice and receipt, available on the Billing page.