AI token & prompt cost calculator

Estimate tokens and input cost for GPT-4o, Claude 3.5 Sonnet and DeepSeek V3 from prompt length. Heuristic only — not a tokenizer API.

Three builds ship from this site. SEO Note 2.1 is the current one: an Electron app (about 121 MB installer) for Windows 10/11 64-bit and Linux x64, with Markdown, the tree editor, AI talking to a model you connect — a local one or any OpenAI-compatible endpoint, a web clipper and git sync of the file. For Android 6.0 and newer there is a release-signed APK (~10 MB) of the same interface; on the phone notes live as a folder of Markdown files on the device, and passwords, sync, the clipper and AI answer there with an explicit refusal. The classic build 1.3.34 (~2.5 MB, Win32) stays on the download page for Windows 7, 8 and 32-bit PCs, where 2.1 does not install; its notes file is .seo. All three work offline with no account: pick your build, or read how 2.1 and 1.3.34 differ.

0

Estimated tokens

$0.0000

GPT-4o Input ($2.5/1M)

$0.0000

Claude 3.5 Sonnet ($3.0/1M)

$0.0000

DeepSeek V3 ($0.27/1M)

Build a prompt library: store reusable prompts and contexts in the free Windows outliner SEO Note — offline, no subscription. Download free (~2.5 MB).

Frequently asked questions

How does this calculator count tokens? The script on the page splits the pasted text into three character groups — Cyrillic, Latin and everything else (spaces, digits, punctuation, non-Latin symbols) — and applies one formula: Cyrillic / 1.7 + Latin / 4 + other / 3, rounded up. No real tokenizer is called; it is arithmetic over text length. The numbers refresh on every edit, whether you type or paste, and an empty or whitespace-only field resets all four cards to zero.

How accurate is the token estimate? It is a length-based estimate, which the card label already says: "Estimated tokens". Every model ships its own tokenizer and returns a different count for the same text, so the provider figure can differ noticeably. Read the result as an order of magnitude, not as a line on your provider invoice.

How is the dollar cost calculated, and which prices are used? The estimated token count is multiplied by the fixed per-million prices printed in the card labels: GPT-4o Input $2.5 per 1M, Claude 3.5 Sonnet $3.0 per 1M, DeepSeek V3 $0.27 per 1M. Those prices are hard-coded in the page and never fetched from a provider, so verify current rates on the vendor site. The figure shows five decimals (six for DeepSeek) because a short prompt costs a fraction of a cent.

Does it include the cost of the answer the model writes back? No, only the input side: all three prices apply to the same input token count, and the card is labeled explicitly "GPT-4o Input ($2.5/1M)". Output tokens, system prompts, chat history, images and cached-input discounts are not modeled. There is a single text field, so two prompt variants cannot be compared side by side.

Which models are available, and can I add my own? Exactly three: GPT-4o, Claude 3.5 Sonnet and DeepSeek V3, with names and prices built into the page script. There is no settings form, so you cannot pick another model, edit a price, switch currency or save a preset. For any other model, take the estimated token count and multiply by its per-million price yourself.

Is my text sent to a server, and do I need an internet connection? No. A JavaScript snippet does the math in your browser: the page has no form and no fetch or XMLHttpRequest calls, so the pasted text is neither uploaded nor stored. You need a connection once, to load the page; recalculation itself is local. There is no Calculate button either — results update as you type. Nothing persists: after a reload the field is empty, there is no calculation history, no account and no sign-up.

Related: AI prompts in notes · Token cost of note workflows · Word counter · Download SEO Note