Glossary · 1 · Foundations: how AI works

Parameters and weights

Also known as: Model parameters, Model weights

Parameters are the numeric values inside a neural network — mostly the weights of its connections — that are learned during training and store what the model knows. Model size is usually given as the number of parameters, from a few billion to more than a trillion.

Intermediate
Technical marketers
Developers

In one sentence

Parameters and weights explained: the learned numbers inside an AI model, what “7B” or “70B” means, and why size isn’t everything.

Example

A model labeled “8B” has about eight billion parameters and can run on a strong laptop; a model with hundreds of billions needs data-center hardware.

Why it matters on your learning path

Technical marketers: Parameter counts are popular in launch announcements but are a weak proxy for quality; benchmark and task results say more.
Developers: Parameter count drives memory and hardware needs. Quantization reduces precision to make large models fit smaller hardware.