Context window checker

Paste text or import a page into this context window checker to see which AI models can read it whole, how much room is left for the answer, and how much to cut when it does not fit.

Free, no sign-up. Your text is counted in your browser, with no AI calls.

Import a public page URL

Only the URL is sent to our server when you import. Pasted text stays in your browser. We read the fetched HTML without running page scripts.

Your text stays in your browser and is never uploaded.

The answer shares the context window with your text. Set the longest answer you expect.

Will it fit?

Paste text or import a page to check it.

Characters
0
Words
0
GPT-5 tokens, exact
0

A context window is the most a model can read at once, counted in tokens, and the answer shares it. GPT-5.4 mini has a 400,000 token window but takes at most 272,000 input tokens. This checker sets your text against each model’s window and input cap, keeps room for the answer and shows what to cut.

How to use this tool

Your text
Paste text, or import a public page by its URL. A page is reduced to its readable text before counting: tags, scripts and styles are removed. Only the URL goes to our server, which checks robots.txt and reads at most 256 KiB of the page.
Token counts
GPT-5 models get an exact count from the o200k encoding, and GPT-6 shows the same count marked not confirmed. Claude, Gemini and Mistral use their makers’ published rules of thumb. Makers with no published rule use the high end of a range, so a fit is never claimed on the low estimate.
Room for the answer
The answer is written into the same window, so the room you set comes off first. Where the model list records a separate input cap, the lower limit applies. A warning shows when your answer room is above the model’s output cap.
The model list
Windows, caps and prices come from the same model list as the model pages and the agent cost calculator. Models a maker has replaced stay hidden until you ask for them.
What to cut
For each model the text does not fit, the table shows the tokens over and about how many words that is, or how many parts to split it into. The longest paragraphs are listed first, since cutting one of them saves the most.

Anthropic: how context windows work

Questions before you start

The most text a model can take in one request, counted in tokens. It holds your instructions, the text you send, earlier messages in a chat and the answer the model writes.

The model writes its answer into the same window. If your text fills it, the request fails or the answer stops short, depending on the model and its settings.

Only for OpenAI’s GPT-5 models and gpt-oss, whose encoding is published. The other rows are labelled estimates from each maker’s rule of thumb. When a fit is close, check it with the maker’s own token count before you send a long job.

Yes. Open the import panel, enter a public URL and the page’s readable text is counted. Pages behind a login, blocked by robots.txt or built entirely by scripts cannot be read.

Pasted text is counted in your browser and never sent. When you import a page, only its URL goes to our server, which fetches the page and returns its HTML to your browser.

Related tools

  • AI token counter

    Paste text into this AI token counter to see characters, words and tokens side by side, with exact counts for OpenAI’s GPT-5 models, labelled estimates for Claude, Gemini, Mistral and Llama, and the cost of each request.

  • Agent cost calculator

    Use this AI agent cost calculator to price conversations from steps, tokens, cache hits and extra billed calls. Add tools and seats, compare annual budgets, and copy the breakdown.

  • ChatGPT prompt generator

    Use this ChatGPT prompt generator to build a complete business brief. Choose ChatGPT, Claude or Gemini formatting, then edit and copy your prompt with a checklist of details to change.

See how AI assistants describe your business

Run the free visibility scan to check which sources assistants use. If you need regular monitoring, compare the plans before choosing one.

Run the free AI visibility scan