How accurate is the token count?

It uses a character-based heuristic (roughly four characters per token for English) that lands within a few percent of OpenAI's tiktoken for typical prose. Code, non-Latin scripts, and heavy punctuation tokenise differently, so treat the count as a close estimate, not an exact match to a specific tokenizer.

Does this count output tokens too?

This tool prices the input (your pasted text). Output usually costs more per token and depends on how long the response is. For a full input-plus-output budget, use the LLM API cost calculator and add an expected completion length.

Where does the pricing come from?

Prices are representative public per-million-token list prices for each model's input tokens. Vendors change pricing periodically and offer batch or volume discounts, so confirm against the provider's current pricing page before committing budget.

Is my text sent anywhere?

No. Token estimation and cost maths run entirely in your browser. Your pasted text never leaves the page.

Token Cost Estimator

Token cost estimator

Before you run a prompt ten thousand times, it helps to know what it costs once. This estimator counts the tokens in any text you paste and prices that input across the major models side by side, so you can see at a glance that the same document might cost ten times more on a frontier model than on a small one — and decide whether the quality difference is worth it.

How it works

The tool estimates tokens with a character-based heuristic (about four characters per token for English text, with a small adjustment for whitespace and punctuation) that tracks OpenAI’s tiktoken closely for ordinary prose. It then multiplies that token count by each model’s input price per million tokens and by your number of requests to give a per-model total. Everything is computed locally; your text is never uploaded.

Tips and notes

For a realistic budget, remember this prices input only — completions usually cost more per token, so add an expected output length on top (the LLM API cost calculator handles both). Token counts vary by tokenizer: code and non-English scripts pack fewer characters per token, so the estimate runs slightly low for them. If your prompt is mostly a fixed instruction with a small variable part, prompt caching can cut repeated-input cost dramatically — worth checking your provider’s caching support. Always confirm the live price on the vendor’s pricing page before sizing a large run.