Short answer: A context window is how much an AI model can read and keep in mind at once — your messages, its replies and any files you attach — measured in tokens. Adral’s general model has a published context window of 1M tokens, roughly 750,000 words of text.
Every AI model works within a limit on how much it can consider at one time. Understanding that limit explains why long projects behave the way they do — and how to get better results when you work with large documents.
First, what is a token?
Models do not read words; they read tokens — small pieces of text. A common short word is usually one token. Longer or rarer words, numbers and punctuation are often split into several.

A useful rule of thumb for English is that a token is about three-quarters of a word. It is an estimate — the exact split depends on the model and the language — but it is close enough for planning.
What the context window is
The context window is the model’s working memory for a task. Everything inside it is read together: the instructions you gave at the start, the files you attached, the replies so far and your latest message. Anything outside it simply is not part of what the model is considering.
- Your messages and Adral’s replies count towards it as the conversation grows.
- Attached files count too — a long contract takes far more room than a short note.
- Bigger is not the same as better reading, but it means less has to be left out.
What 1M tokens means in practice
Adral’s general model has a published context of 1M tokens. As a rough guide, that is about 750,000 words — on the order of 1,500 pages of text — in a single context.
| Material (rough estimates) | Approx. tokens | Share of 1M |
|---|---|---|
| A 4-page project brief | ≈ 2,500 | < 1% |
| A 180-page contract | ≈ 110,000 | ≈ 11% |
| 60 pages of research notes | ≈ 36,000 | ≈ 4% |
| A full novel (~90,000 words) | ≈ 120,000 | ≈ 12% |
Estimates assume about ¾ of a word per token and about 650 tokens per page; real numbers vary with layout, language and how the model splits the text.
How to work well with a large context
- Put the brief first. State what you want, what it should contain and what “done” looks like before the long material — see Writing a good brief.
- Point at what matters. “Compare section 7 of the contract with the Q2 sales sheet” beats “look at everything”.
- Start fresh for a new task. A new conversation keeps unrelated material out of the way.
- Still check the output. A large context lets Adral read more; it does not replace your review — see Checking AI output.
Frequently asked questions
What is a context window in simple terms?
It is how much text an AI model can read and keep in mind at once — your messages, its replies and any attached files — measured in tokens.
How big is Adral’s context window?
Adral’s general model has a published context window of 1M tokens, which is roughly 750,000 words of text.
How many words is a token?
In English, a token is roughly three-quarters of a word on average. Short common words are usually one token; long or rare words split into several.
Do attached files use up the context window?
Yes. Everything the model reads for the task counts — including attached documents, which is why a 180-page contract uses far more than a short note.
Related: The Adral models · Writing a good brief · Checking AI output



