Cost per token

The posted price of input and output tokens consumed or generated.

Also called

Inference The act of running a trained model to produce an output from inputs, as opposed to training it.
Batch inference A type of inference where many requests are sent together and answered later, usually at a lower price.