Foundation Model

Definitions

1. A large AI model pre-trained on broad data that can be adapted to a wide range of downstream tasks through fine-tuning or prompting. GPT-4, Claude, Llama — these are foundation models. Stanford coined the term to emphasize that these models serve as the foundation for many applications.

   In plain English: A powerful AI model trained on massive amounts of data that can be specialized for many different tasks — like a well-educated generalist who can learn any specialty quickly.

2. The economics of foundation models create a natural oligopoly: training costs hundreds of millions of dollars, requiring massive capital and compute. A handful of labs (OpenAI, Anthropic, Google, Meta) train the foundation models; thousands of companies fine-tune and deploy them.

   Example: 'We don't train our own foundation model — that's a $100M endeavor. We fine-tune Claude for our domain. The foundation model is the platform; our data is the differentiator.'

   Source: economics / industry structure

Related Terms

• Large Language Model
• Multimodal
• Weights
• Open Weight