Back to Lexicon
Wild WordAI Experience

context window

How much text a model can see at once in a single session — the working memory of the conversation.

First Seen

2017

Context

Measured in tokens. Early GPT models had 2K–4K; modern models reach 128K–1M+. A larger window allows more document context, longer conversations, and complex multi-file reasoning. Distinct from 'context rot,' which describes quality decay inside the window as it fills.

Citation

Transformer Architecture / LLM Documentation

2023

https://en.wikipedia.org/wiki/Large_language_model

A wild word — documented from usage in the field, not coined by Xenolexica.

Understanding & Perception