inference
Using a trained model to get an answer, as opposed to training it.
Inference is the everyday work a model does after training: taking a new input and producing an answer. Every question to a chat assistant and every photo checked by an app is one inference. Training happens once, but inference happens every time someone uses the model. So its speed and cost matter most day to day.
Think of it as
Like a trained chef cooking tonight's orders. Culinary school was the training; each plate that leaves the kitchen is an inference.
Example
A shop's product-photo checker was trained once last year. Every time a seller uploads a new photo, the model runs an inference to decide if the photo meets the rules. Ten thousand uploads a day means ten thousand inferences a day.