AI Token Counter — Estimate Tokens & API Cost Before You Pay

How to use

1. Paste your text — see instant character, word and estimated token counts.
2. Set the rates — enter the model's $ per 1M input and output tokens (defaults provided, always editable).
3. Read the cost — estimated input cost, and projected total if you also set expected output tokens.

About this tool

LLMs read text as tokens — roughly one token per 4 characters or ¾ of a word in English. Providers bill per million tokens with separate input and output rates. This counter estimates tokens from your text using the standard ≈chars/4 heuristic, then multiplies by editable per-model rates — update the two numbers whenever pricing changes and the cost math stays honest. Nothing is sent anywhere. Every AI token calculator walks a line: exact counts require each provider's own tokenizer, while this tool uses the industry-standard heuristic — roughly one token per 4 characters or ¾ of a word in English — which lands within a few percent for typical prose and is instantly good enough for the real question: what will this prompt cost. The pricing side is deliberately manual. Providers reshuffle per-million rates and add cache discounts often enough that any hard-coded price table eventually lies to you; entering the two numbers from the provider's current pricing page keeps the math permanently honest. Remember the asymmetry that surprises everyone: input tokens are what you send, output tokens are what the model generates — and output rates are usually several times higher, so a chatty model on a cheap input can still be an expensive call. Context limits count both together.

FAQ

Rate this tool

Your rating stays in this browser (local storage) — nothing is sent to any server.