A token limit is the maximum amount of text an AI model can process or generate in a single request, measured in tokens rather than words. A token may be a whole word, part of a word, punctuation, or a symbol, so token counts do not map cleanly to plain word counts.
In cybersecurity, token limits matter because they affect how much context a model can inspect and how much output it can safely produce. Long incident reports, source code, logs, and threat intel can exceed the limit and get truncated, which may hide important clues. Attackers can also abuse large contexts by burying prompt-injection text inside files or tickets that a model reads. Defenders manage this risk with chunking, retrieval, truncation checks, prompt filtering, and strict access controls so the model only sees what it needs.