Every AI API price change, on the record.
We read the published pricing pages of every major model provider every three hours and write down what moved. Launches, cuts, quiet increases, deprecations, retirements — timestamped, sourced, kept forever.
Gemini 3.6 Flash goes up 100%.
Introductory pricing ends; Google has published the standard rate that takes effect after it.. Per million tokens. Recorded from Google's own pricing page.
Get told before the next one →Latest changes
Full change log →Cheapest models in general availability
All 168 models →Ranked by a blended cost of one million input tokens plus three million output tokens — a rough stand-in for a real chat workload.
Stop hardcoding a table that rots.
Every app that touches an LLM ends up with a pasted map of model ids to prices and context windows, and it is wrong within a month. Fetch the live one instead — open data, no key, MIT licensed.
Read the API docs Package on GitHub
Works with any coding agent over MCP, so it quotes today's prices instead of guessing from training data.
import { costOf, scheduledChanges } from 'statlyte';
// npm i statlyte
await costOf('claude-sonnet-5', {
input: 12_000, output: 800
});
// => 0.032
await scheduledChanges();
// [] today — populated the moment a vendor announces
// a future increase, e.g. { name: 'Claude Sonnet 5',
// effectiveOn: '2026-09-01', from: {...}, to: {...} }
Know before the finance review.
Watchlists per environment, an alert when a model you bill against moves, and a permanent record you can point at.
Cut spend without guessing.
Price your real monthly volume against all 149 models in general availability and see what a switch actually saves.
A citable price history.
434 timestamped observations, each with the vendor page it came from. Export as CSV or pull it from the API.