Free tool
What does your document
cost an AI to read?
Paste text or set a page count, pick a context window, and see the token bill format by format: scanned PDF versus Word versus structured note. The differences are not small.
Nothing you paste leaves this page.
Used only when the text box is empty. Think of a typical report page.
Check your AI provider's current pricing; this powers the cost column.
As a structured note, this document costs
–
Format by format
| Format | Tokens | Of window | Cost / read |
|---|
A scanned PDF costs about – the tokens of the same content as a structured note.
Why token counts decide what your AI can know
Every AI model reads through a context window measured in tokens, and every token costs money and latency. When your knowledge lives in scanned PDFs, most of the window is spent on format overhead and OCR noise. When it lives in structured notes, the same knowledge fits in a fraction of the space, which means an agent can read ten documents instead of one before it answers.
What this means for a knowledge base
This is a core reason NoteX stores notes as structured text files. AI add-ons on the desktop and agents connected through the API and MCP read .notex and Markdown content directly, with no conversion step and no OCR pass. See speed and API & MCP for how that plays out in practice.
Counting tokens for a real project? Paste representative text above for an exact character-based estimate rather than the page model, and check your provider's tokenizer for the final word.
Frequently asked questions
How accurate are these numbers?
They are illustrative, based on our internal benchmarks of typical documents (roughly 4 characters per token for text, with per-page overhead factors for each format). Real counts vary with content, tokenizer, and OCR quality, but the relative gap between formats is robust.
Why do scanned PDFs cost so much?
A scanned page is an image. Reading it means OCR output plus layout noise, or vision-model processing, both of which multiply tokens. The same paragraph as plain structured text is an order of magnitude smaller.
What counts as a structured note?
Markdown, .notex files, and similar text-first formats where the structure is lightweight markup rather than a binary container. This is the format NoteX stores everything in.
Does the calculator send my text to a tokenizer API?
No. Everything runs in your browser with a character-based estimate. For exact counts, run your provider's tokenizer on the final text.
Your notes deserve a better engine.
A Notion-style block editor on the engine behind the world's most popular code editor. Notes stay files you own, extensions add what you need, and AI plugs in natively. Free to start, on web and desktop.
Free to start · Your files, your disk · Windows desktop & web