The Definitive Guide toAI Data Centers
Ask the Guide
GuideGlossaryPrefill

Prefill

The compute-heavy phase that processes an inference prompt and builds its KV cache before generation begins.

← All terms