Pricing model
Buy credits once. Spend them against a published model price.
Consumers fund a platform service balance, then choose which callable model each key may use. Token prices come from an immutable price version; the platform keeps 10% of confirmed usage and the serving supplier earns 90%.
Consumer pays
100%
The confirmed platform service price for actual usage on the selected model.
Serving supplier earns
90%
Pending earnings mature after the configured risk period and payment availability checks.
Platform service fee
10%
Payment fees and infrastructure are borne by the platform under the current design.
Model token prices
Browse every published model, compare input / output / cache rates, and open a detail page for curl, JavaScript, and Python call examples. Preview rows are labeled and cannot be purchased yet.
Credits are a liability until consumed
A top-up does not create supplier earnings. Only a request that names an allowed model and settles verified usage produces the 90/10 split. Reservations protect concurrent budgets but do not recognize revenue.