API pricing and balance: how requests are billed

On the API you pay only for what you use: no monthly plan and no minimum number of requests. Each request is paid from your balance according to the model and the number of tokens, and every model costs less than its official price.

How a request is priced#

• Prices are per million tokens, with one price for input (what you send) and one for output (what the model writes).
• A request costs its input tokens plus its output tokens, at the price of the model you called.
• The token counts come back in the reply itself, in the usage field.

An example with the default model:

Our prices follow the published official prices, and the date they were last checked shows with every price.

Model prices#

The full list, including image and embedding models, is on the models page.

Your balance#

• You top up from your dashboard with one of the payment methods; in Egypt, in Egyptian pounds.
• Before you pay, the top-up screen shows the amount you pay and the balance that will be added to your account.
• There is one balance for your whole account, and the requests of all your keys draw on it.
• If your balance doesn't cover a request, you get INSUFFICIENT_BALANCE.

Track your usage#

Your dashboard shows your usage and its cost per key and per model, with a log of requests. To cap a single project, set a spend limit on its key in the key settings.

What about subscriptions?#

Subscriptions are a different path from the API: a fixed monthly price for the official apps, for less than their official price, with two usage limits only: a 5-hour limit and a weekly limit. The plans and their prices are on the pricing page.