Context Window
The context window is the model's working memory, measured in tokens: everything in a single prompt (instructions, documents, the conversation so far) must fit inside it, and anything beyond the limit is simply invisible to the model.
No jargon, no math
In plain words, what is it?
The context window is the model's desk: everything it can see at once, measured in tokens. Your instructions, the pasted documents and the conversation so far all share this one space. Inside the window the model can connect anything to anything; outside it, nothing exists. That gives you three hard rules. First, size: a window of 128,000 tokens sounds huge until a 200-page contract eats it, and past the limit you get an error or a silent truncation, not a polite warning. Second, memory: the window resets between separate calls unless your system re-sends the history, which is why the model forgets Monday's email on Tuesday. Third, attention: models are sharpest on what sits at the start and end of a long document, and noticeably weaker on what hides in the middle, the documented 'lost in the middle' effect. Working around the window is a discipline: chunk long material, summarise hierarchically, retrieve only what matters. RAG exists largely because of this limit.
Terms that live nearby.
Token
A token is the chunk of text a language model actually reads and writes: on average about three quarters of an English word, converted to a number from a fixed vocabulary before the model can process it.
ReadTokenization
Tokenization is the step that turns text into tokens, the small units a language model actually reads: the input is split according to a fixed vocabulary and each token is mapped to a number, so the model processes IDs instead of words.
ReadLarge Language Model
A large language model (LLM) is an AI system trained on massive collections of text to predict the next token, which gives it a general ability to write, summarise, translate and reason without being programmed for each task.
Read§ 03
Want this running in your stack?
We design and ship AI systems with permissions, logs and human approval. It starts with a 30-minute scoping call.