Inference

Running a trained model to produce output - what happens every time you send a prompt.

Training builds the model; inference uses it. Every request you send is an inference run, and its cost is dominated by how many tokens go in and come out.

Two phases matter for latency. Prefill processes your whole prompt at once and is fast but grows with prompt length. Decode generates the answer one token at a time and sets how long you wait for a long response. That is why time-to-first-token and total time behave differently, and why streaming makes an answer feel faster without being faster.

Tagged

Where this comes up

Prompts, configs and tutorials in the library that touch this term.