Inference
Running a trained model to get an answer. Everything you pay for by the token is inference, not training.
Also written: inference time, serving.
Training is building the model. Inference is using it. They are separate activities with wildly different costs: training a frontier model (The largest and most capable model a vendor currently offers, priced accordingly.) costs tens or hundreds of millions of dollars once, and inference costs fractions of a cent, every time, forever.
Everything on this site is inference pricing. You are never charged for training, and no amount of your usage teaches the model anything — the weights are fixed between releases.