Context Window

The context window is the total amount of information a language model can consider when generating a response — the user's prompt, conversation history, retrieved documents, tool outputs, and any system instructions, all counted together. Everything outside the window is invisible to the model. Larger windows let models consider more material in a single request, which helps with tasks like reviewing long contracts or synthesizing multi-document research. But a larger window doesn't guarantee better answers — models can still overlook, misprioritize, or misinterpret information even when it fits.

Context window size shapes what's architecturally possible with AI tools. An assistant that silently truncates a long document because it exceeds the window won't surface an error — it will just miss whatever was cut. This affects how enterprise AI tools are designed: when to use retrieval to fetch only the relevant sections, when to summarize before passing material to the model, and how to test that key content isn't being dropped. Cost and latency also scale with context size, which affects how systems are optimized for production use.