Context Window
The maximum amount of text (in tokens) a model can read at once.
A context window is the token budget a model can ingest in a single request. 100K tokens is roughly 75K words — enough for a full book chapter. Claude Opus 4.6 now supports a 1M-token window, letting it reason across hundreds of files. Bigger isn't always better, but for large codebases it's usually necessary.