Foundation Model is a large AI model, trained on a broad range of data, that can be adapted to a wide variety of tasks. Large language models like GPT are the best-known examples.
What it means in plain English
A foundation model is a single, general-purpose model that serves as the base for many applications. Rather than training a separate model for each task, developers start from a foundation model and adapt it — through prompting or fine-tuning — to their specific need. Its broad training gives it wide-ranging capabilities out of the box, which is why one model can write, summarise, translate, and code.
The term captures a major shift: from many narrow models to a few powerful, adaptable ones.
A simple example
A large language model is a foundation model: the same base model powers a customer-support bot, a coding assistant, and a writing tool, each adapted from the same foundation rather than built from scratch.
Why it matters
Foundation models represent the dominant paradigm in modern AI. Their generality is why AI capabilities have spread so fast — a single model can be pointed at countless tasks, and improvements to the foundation ripple out to everything built on it.
Related terms
- Large Language Model — the most common kind of foundation model.
- Pre-training — how foundation models gain broad capability.
- Fine-Tuning — how they are adapted to specific tasks.
Frequently asked questions
What is a foundation model?
It is a large model trained on broad data that can be adapted to many downstream tasks — large language models and major image models are examples.
Why are they called “foundation” models?
Because they serve as a base that many applications build on, through prompting or fine-tuning, rather than being trained from scratch for each task.
