In short
- The price per million tokens is the standard way APIs are priced: separately for input and output tokens. You use it to calculate a task's cost.
Makers give two prices: per million input tokens (what you send) and output tokens (what the model generates). The cost of a call is input tokens divided by a million and multiplied by the input price plus output tokens divided by a million and multiplied by the output price. At makers who bill reasoning, thinking tokens are counted in the output too. Caching and batch processing have separate, lower prices.
A caution
Prices have a date and are sometimes announced with a schedule, so record the reading date and the page address. Compare the cost of a task, not table prices, as in the article Comparing AI model prices.
Where to read next
Task cost in tokens. The concept token. The whole topic: the Model and API pricing hub.