Context window
Also called context length
How much text a model can take into account at once: the chat so far, files and instructions together.
Everything the model considers when it writes a reply has to fit in its context window: the app's hidden instructions, your memory notes, the whole conversation so far, attached files and search results. The model rereads all of it every turn. It has no other short-term memory.
2 things go wrong as a chat grows. Once the window is full, the oldest parts are dropped or summarised, so early decisions quietly disappear. And well before it's full, models handle long input less reliably: details in the middle of a long chat or document get missed. A short handover into a fresh chat often beats patching a long one.
For example
You set a budget of £400 at the start of a long trip-planning chat; 40 messages later the model suggests a £650 hotel, because that early message is slipping out of view.
Put it to use
Where you'll meet context window in ChatGPT, Gemini and Claude, how to use it there, and the mix-up most people make.
Daily Bytes Plus is in the Android app.