Loading…
Glossary term
The maximum amount of text, measured in tokens, a model can consider at once.
The context window is everything the model can see at once: the system prompt, your conversation, any documents and its own reply so far. It is measured in tokens, and anything beyond it is simply not visible to the model.
Very large windows now reach hundreds of thousands to millions of tokens, but a model may not use all of it equally well.
It limits how much a model can read or remember in one go, and long contexts cost more to process.