A context window is the maximum amount of text, measured in tokens, that a language model can take into account at once, spanning both the input it is given and the output it generates. Anything outside this window is invisible to the model unless it is retrieved and re-supplied.
Why it matters
The context window sets a hard limit on how much a model can "see" in a single pass, which is why long documents and long histories must be chunked, summarized, or retrieved selectively. Even as windows have grown to a million tokens or more, effective use degrades on very long inputs, with models tending to lose information buried in the middle. This is the core constraint that makes retrieval and external memory necessary rather than optional.
How Pith relates
Because no context window can hold everything you have ever read, Pith keeps your knowledge in a searchable, source-grounded store and surfaces only the relevant, cited passages when an AI assistant needs them. This sidesteps the "lost in the middle" problem by feeding focused, attributable context instead of dumping in raw history. The result is grounded answers that do not depend on stuffing everything into one prompt.
See also
Last reviewed: 7 June 2026 · Licensed CC BY 4.0 · cite freely with attribution to Pith.