About
A ledger for people who pay per token
TokenLedger is a free, single-purpose cost calculator for teams that build with language models.
Why it exists
Most teams discover their token bill twice: once when they estimate it from a prompt length, and again when the invoice arrives. The gap is rarely a pricing mystery. It is output tokens billed at four to ten times the input rate, a system prompt resent on every call, a cache that never hits because a timestamp sits at the top of the prefix, or a model tier that was chosen in a hurry.
This site puts those numbers on one screen before the invoice exists. Paste the prompt, set the workload, compare the models, find the crossover with seat pricing, and export the result with its assumptions attached.
What it is not
It is not a model playground, an LLM proxy, or a benchmark. It never calls a model, never stores your text and never asks for an account. It is also not a promise: token counts for Claude and Gemini are estimates because their tokenizers are not published, and no calculator can see retries, reasoning tokens or the price change scheduled for next quarter.
How the numbers are kept honest
- Prices come from each provider's published pricing page and carry the date they were read. The current page was checked on 18 November 2026.
- OpenAI token counts use the real o200k_base vocabulary, run locally. The estimator's error was measured against that vocabulary rather than asserted.
- Every calculation is visible in the page's JavaScript, and the exported files state the workload they assumed.
- Corrections are welcome and are the fastest way to improve the site: write to [email protected] with the model, the price you see and a link to the provider's page.
Who it is for
Engineers choosing a model, founders pricing a feature, finance teams checking a forecast, and anyone who has ever multiplied a token count by a price per million and wondered whether the answer was plausible. The calculator is free and needs no sign-up; advertising keeps it that way.