Token Bill is an independent, free tool that shows what an AI feature will cost before you build it.
You enter how many requests you expect, how long the prompts and answers are, and how much of the input repeats. The calculator prices the same workload on every model in the list, side by side, with long-context surcharges, batch discounts and cached-read prices.
The Which model? pages shortlist models for common jobs, and the Subscription or API? page compares chat plans with paying per token.
Prices are read from each provider's official pricing page, linked at the bottom of the calculator, together with the date of the last check. Prices for open models such as Llama and Qwen come from the third-party host named next to the model. Plan prices are shown as the provider publishes them.
Prices change often. Confirm on the official page before you commit to a budget.
Token Bill estimates cost. It does not measure how good a model is on your task: the quality columns come from public leaderboards and are a rough guide only. Estimates leave out taxes, cache write and storage fees, tools and search, image and audio pricing, and hosting. Test the shortlisted models on your own examples.
Token Bill is not owned by, or affiliated with, any AI provider. No provider pays to be listed or ranked today: models are ordered by calculated cost. If we add affiliate links or advertising later, we will label them clearly.
Found a wrong price or spec? Write to [email protected] with the model name and a link to the official page, and we will check it.