Context window
The most tokens a model can take into account at once, covering the prompt, any documents and its reply.
Also: context length
The context window is the amount of text a model can consider in one go, measured in tokens. Early chat models handled a few thousand; by 2025 many offered 128,000 to a million or more. Longer contexts let a model read whole contracts or codebases, but they cost more: the prefill takes longer and the KV cache uses more GPU memory per user. Providers often price long-context requests higher for this reason.