Context window

Also: context length

The maximum number of tokens a model can attend to in one call — system prompt, retrieved material, conversation history and the reply being generated, all counted together.

What people get wrong

Advertised length is not usable length. Accuracy on material buried in the middle of a very long input degrades well before the stated limit, and cost and latency rise the whole way there.

A definition is not an understanding. This is taught properly in The window as a budget, part of Context windows.

Learn this properly