Foundation model
A big, general model trained once on broad data, then adapted to countless specific jobs instead of building one from scratch each time.
Think of it like
A liberal-arts graduate you can train up for almost any role, versus hiring a narrow specialist per task.
Example
The same base model powers a coding assistant, a support bot, and a summarizer.
How it actually works
The term captures a shift: instead of training a bespoke model per task, you train one large general model and adapt it — by prompting, retrieval, or fine-tuning. That reuse is why capabilities spread so fast, and why a handful of expensive base models sit under thousands of products.
For product teams
Build on one general model and specialize it — cheaper and faster than bespoke models per feature.
For engineers
A large pretrained model reused across tasks via prompting, RAG, or fine-tuning.
Related
- LLM — An LLM is a text-based foundation model.
- Fine-tuning — Adapted to a niche via fine-tuning.
- Prompt engineering — Or steered at run time with prompt engineering.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome