FAQ
Product
What is this?
A public board of LLM list prices, and a free program you run locally. The program quotes those prices for a token count and can mask emails, phone numbers, and IPv4 addresses with regular expressions.
Does it save 80%?
No. The site will show the arithmetic difference between a flagship list price and a cheaper model's list price. That is not a measured saving, and it says nothing about answer quality.
Does the quote demo call the model it names?
No. The quote page estimates tokens in your browser and multiplies by the catalog. It does not send the prompt to Groq, OpenAI, Anthropic, or anyone else.
Do I need a license?
No. Download the zip and install it. Checkout is closed.
Prices
Which prices are these?
Short-context public list prices, in USD per 1,000,000 tokens. Cached input, long context, and batch rates are called out in notes when we track them. They are not included in the headline number.
How often do they change?
The catalog is refreshed on the 1st of each month from the vendor pages linked on the calculator. A long-running server also refreshes in memory when the stamped date is 30 days old.
Why are some Claude names dotted?
The page shows claude-haiku-5.5. The API id, when you call Anthropic yourself, is claude-haiku-5-5. Same pattern for Fable, Opus, and Sonnet.
Redaction
What does anonymize catch?
Emails, phone numbers, and IPv4 addresses. Long digit runs only if you pass --numbers. It does not find names.
Is this enough for GDPR?
No. It is a local regex pass. Read the output, and get advice for your own compliance work.
Does the redaction demo upload my text?
The browser demo runs the patterns locally. The CLI does the same on your machine.