Tools
Small calculators for the questions that come up when you are costing an AI feature. Everything runs in your browser — nothing you type is uploaded, logged or stored.
Which tool answers which question
| Question | Tool |
|---|---|
| How many tokens is this prompt, document or chat history? | Token counter |
| What will a feature cost per month on a metered API? | LLM cost calculator |
| At what volume does a flat-priced instance beat per-token billing? | LLM cost calculator |
How the numbers are produced
The token counter is an estimator and says so: it applies the usual ratios of roughly four characters per token for English prose, with more tokens per word for code, markup and most other languages, and it shows a range. Exact counts depend on each model's tokenizer, so treat the figure as a planning number and use your provider's tokenizer when you need the exact count.
The cost calculator ships with September 2026 list prices as editable defaults. Providers change prices often, so every price field can be overwritten; the value of the tool is the structure of the comparison — metered API against a flat instance — rather than a figure to the cent.
For the reasoning behind the numbers, the guide to cutting an LLM bill covers caching, routing and batching, and running an LLM on CPU covers what a flat-priced instance can realistically serve.
Put your AI stack on your own box
Pick an app, pick a size, and have it running today. Month to month, cancel whenever.