What is a Context window?
A context window is the maximum amount of text, measured in tokens, that a language model can consider at once, including instructions, documents, conversation history and its own reply.
In more detail
Windows now run from tens of thousands to over a million tokens, depending on the model. A token is roughly three-quarters of an English word. Bigger is not automatically better. Cost and latency rise with input size, and models can miss details buried in the middle of very long inputs, so production systems usually pass selected excerpts (through retrieval) rather than everything. Conversations also grow. After enough turns, early messages fall out or must be summarized, so plan for that. If a rule must always apply, put it in the system prompt on every call instead of trusting it to survive a long thread.
Example
A 300-page lease will not fit in a small window. The system extracts the clause list first, then sends only the rent review and break clauses to the model that answers a tenant's question.