Model
The "brain file" an AI produces after learning. You give it an input and it gives back an answer.
In everyday terms
A model is the finished result of training: a big file of numbers that turns inputs (a photo, a question) into outputs (a label, a reply). GPT, Claude and Gemini are models.
For professionals
A parameterised function f(x; θ) whose parameters θ were fitted during training. Shipped as weights plus an architecture definition.
Think of it like…
A recipe that was perfected by trial and error. Once written down, anyone can cook from it.
You've already seen it
When an app says "powered by GPT-5" or "using Claude", that's the model.
Myth vs reality
Myth: The model looks things up in a database.
Reality: A model stores learned patterns as numbers, not a searchable copy of its training data.
Quick check
After training is finished, what do you actually have?
- A database of every example
- A model: learned numbers that map inputs to outputs
- A list of rules
- Nothing permanent
Show answer
A model: learned numbers that map inputs to outputs: Training produces the model's parameters.