PRICING

Pay for the work you use.

Choose a model. See its rates. Start building.

The catalogue.LIVE

Reading the model catalogue…

HOW BILLING WORKS

A rate for your model.
Only the work you use.

Choose your model

Each model has its own rates. Compare input, output, and cached input before you start.

Pay by usage

Token rates are per million tokens. Models with a flat rate are billed per request.

Keep track of your spend

Your workspace shows usage, available credit, and the charges for your requests.

Read the billing guide

Explore request priority Coming later

Your work. Your pace.

A preview of per-request priority: wait longer to pay less, or pay more to go first. All requests currently use the standard rate.

StandardBase rate

First token · TTFT600ms

Comparable standard performance

Generation · TPS50tok/s

Comparable standard throughput

For everyday APIs and interactive apps. Start here.

Per-request priority is not live yet. Every request today is served at the standard rate on the left, which is the price you are charged. The curve and the service figures here describe what we are building, not a rate you can select or a measurement of the fleet.