Inference

Simple

A trained AI at work: the act of producing an answer.

Business

The operational phase of a model, where all request costs live.

Tech

The model’s forward pass in production; metrics: latency, throughput, cost per token.

Related terms

All glossary terms