← Concept Index

Model foundations

Foundation models

Also called: base model, general-purpose model, pretrained model

DEFINITION

A large model trained once on broad data to be a general-purpose base, which many different products then adapt or build on rather than training from scratch.

WHY IT MATTERS

It's why so many tools feel similar: they often sit on the same handful of foundation models. It also means a limitation or bias baked into the base can surface across every product built on it.

COMMONLY CONFUSED WITH

The app you use. The product (with its prompt, guardrails, and data) is built on top of the foundation model; they can behave very differently even when the base is the same.