What is a model?

Model as the engine

A language model reads your text and writes a reply. It does not run on your laptop unless you host one locally; most teams use a provider over the network.

The model name (for example gpt-4o-mini) tells PraisonAI which engine and pricing to use.

Agent wraps the model

An Agent adds instructions and optional tools around that call. Same model, different instructions, different behaviour.

You choose the model; instructions say how it should act.

Example

# Ideas only — nothing runs here
# model = which engine answers
# instructions = how it should behave
# start("...") = your question this turn