A context window is the amount of text an AI model can consider at once, measured in tokens (roughly three-quarters of a word each). It covers everything in a single exchange: your prompt, any documents or history you include, and the model's own reply. Go past the limit and the earliest information falls out of view, so the model effectively forgets it.
Modern models have stretched this limit dramatically, from a few thousand tokens to windows that hold hundreds of thousands, enough for a full brand book, a research deck, and a project's message history at the same time. A larger window means the model can reason over more of your material without you first cutting it down to fragments.
Why it matters for creative teams
For a creative team, the context window is the difference between an assistant that knows your brand and one that needs re-briefing every time. If the model cannot hold the tone-of-voice guide, the client's past feedback, and the current brief together, it produces generic work that misses the specifics. The window sets the ceiling on how much shared context the AI can actually use in one pass.
A real example
Say a strategist asks AI to draft a campaign concept. With a small context window, they can paste in the brief and little else, so the output ignores the brand's rules and last quarter's learnings. With a large window, the same request can include the brand guidelines, the creative brief, the results of the previous campaign, and the client's objections from the last review. The concept that comes back is on-brand and informed, because the model saw the whole picture.
This is why platforms built for creative work focus on getting the right material in front of the model, not just having a big window. A window is capacity. Deciding what earns a place in it is context engineering. The two work together: the window sets how much the model can hold, and good context management decides what fills it.