Foundation models are large, general-purpose AI models trained on massive datasets that serve as a starting point for a wide range of downstream tasks. Rather than building a new model from scratch for every application, organizations take a foundation model and adapt it through fine-tuning, prompting, or retrieval augmentation.
The term reflects the idea that these models provide the foundation on which more specialized capabilities are built. They give every organization — regardless of size — access to powerful, state-of-the-art language capabilities without the prohibitive cost and expertise required to train a model from scratch.
