Context windows: will my text fit?

The context window is the most tokens a model can handle in one request, counting everything.

What is counted

The window covers the instructions, the earlier conversation, the documents you paste, and the answer the model writes. If a window is 128,000 tokens and your input is 120,000, only about 8,000 remain for the answer. The estimator subtracts the reserve you type for the answer before it says whether the text fits.

Checking a document

1. Paste the text. Use the high estimate.
2. Type the window size of your model (from the maker's page).
3. Type the tokens you want to keep free for the answer.
4. Read the result: it shows how much of the available space the text uses.

If it does not fit

• Shorten by removing parts that do not matter to the question.
• Split the document and ask about each part, then combine the answers.
• Summarize first, then ask about the summary.
• Use a model with a larger window, if the cost is acceptable.

Longer is not always better

A long input costs more, takes longer and can make it harder for the model to focus on the part that matters. Give it what it needs.

Words and pages

Using the Gemini rule, 100 tokens is about 60 to 80 English words, so 1,000 tokens is roughly 600 to 800 words. A page of 500 words is roughly 625 to 833 tokens by that rule.

This is an independent guide. It is not connected with Google, Anthropic, OpenAI or any company named here. Facts were read on 14 November 2026 from the pages named on each page. The numbers are estimates from rules of thumb, not the counts of any maker's tokenizer.