How much a model can consider at once.
THE DEFINITION
The amount of text a model can take into account in one exchange: your input, the conversation so far, and its output.
Measured in tokens.
WHY IT MATTERS
Everything the model knows about your situation must fit inside it.
Beyond that limit, earlier content is dropped.
WHAT THIS LOOKS LIKE
An assistant that appears to forget the beginning of a long conversation.
WHAT ELSE GOES WRONG
Attention is uneven across a long context. Material in the middle receives less weight.
WHAT TO DO
Put important instructions first and repeat critical constraints at the end Start a new conversation for a new topic Split very long documents
HOW SIZES HAVE CHANGED
Early models handled a few paragraphs. Current ones handle substantial documents.
RELATED TERMS
Token, conversation, prompt.
