Foundation model
A large model trained on broad data that can be adapted to many tasks. A wider term than language model.
Also: general-purpose AI model
The term foundation model was coined by Stanford's Center for Research on Foundation Models in 2021 for models like BERT, GPT-3 and CLIP that serve as a base for many downstream uses. It covers language, image, speech and multimodal models. Regulators often use similar language: the EU AI Act speaks of general-purpose AI models. Some researchers dislike the term, arguing it implies a solidity these models have not earned; others prefer it because it is not limited to text.