Core concepts
Foundation Model
A large, general-purpose AI model trained on broad data that can be adapted (via prompting or fine-tuning) to many downstream tasks.
01 ——
In plain English
A foundation model is a large AI model trained on a vast and diverse dataset that serves as the base for many specific applications. The term was coined at Stanford in 2021 to describe the new paradigm: instead of training one model per task, you train one big model and adapt it for everything.
Examples of foundation models:
- Language: GPT-4, Claude, Gemini, Llama
- Image: Stable Diffusion, Flux, DALL-E
- Multi-modal: GPT-4o, Claude 3.5, Gemini Ultra
- Code: Codex, DeepSeek-Coder
Why "foundation":
- One model, many uses — chat, coding, summarisation, search, agents
- Adaptation via prompting (no retraining) or fine-tuning
- Often called "frontier models" when they're at the cutting edge
Foundation models are the building blocks underneath nearly every AI tool you'll find in this directory.
02 ——
Related terms
LLM
Large Language Model — the type of AI behind tools like ChatGPT and Claude, trained to understand and generate text.
Generative AI
AI systems that create new content — text, images, audio, video, or code — rather than just classifying or predicting from existing data.
Fine-tuning
Further training a pre-trained AI model on your own data to specialise it for a specific task or style.
Transformer
The neural network architecture introduced in 2017 that powers nearly every modern LLM, image generator, and AI breakthrough.
GPT
Generative Pre-trained Transformer — the architecture behind OpenAI's models, and now used as shorthand for any LLM-powered chatbot.
Back to glossaryLast reviewed May 2026