Knowledge Maxxing← Back to topic library
AI MaxxingBeginner75 sec
Context window
A context window is the maximum amount of text, measured in tokens, that a language model can consider at once when generating a response.
Prerequisite: Token
What matters
Anything beyond the context window is invisible to the model during that request, so long conversations or documents may need to be summarized or split to fit.
Example
A model with an 8,000-token context window cannot directly reference details from page one of a 50,000-token document, unless that section is resupplied in the prompt.
Common misconception
A larger context window does not guarantee the model uses every part of it equally well; information in the middle can still be underused.
Recall question
What happens to information that falls outside a model's context window?
Keywords and related concepts
context windowtoken limitlong inputtoken