About TokenCost

TokenCost is a free token counter and API cost calculator for large language models. It exists because pricing pages are scattered across a dozen vendor sites, quoted per million tokens, and almost never answer the question developers actually have: what will this feature cost me per month?

How token counts work

The counter runs a GPT-4o/GPT-5 compatible byte-pair tokenizer in your browser. For OpenAI models the count matches the API exactly. Anthropic, Google, Meta and Qwen use different tokenizers, so counts for those families typically land within about 10% — close enough for budgeting, not exact for billing.

Where prices come from

Prices are list rates published on each provider's own pricing page, last verified 2026-10-20. They exclude enterprise contracts, batch discounts and negotiated rates. Prompt caching prices are included where the vendor publishes them.

Prices in this market change frequently. If you spot something stale, treat the provider's page as authoritative — and expect this table to lag by at most a couple of weeks.

Privacy

There is no account system, no backend, and no analytics on the text you paste. Tokenization happens in JavaScript on your device. You can load the page, switch off your network, and the calculator keeps working.

How this stays free

The site is supported by lightweight advertising. There are no paywalls, no sign-up walls and no artificial limits on how much text you can analyze.