Nikclas Compares AI Model Prices Before You Build

Summary

Nikclas is an open-source single-page React application designed to answer which AI model can perform a comparable job for the lowest cost. It presents a live price board for four listed flagship models and a head-to-head comparator where users select a provider and model on each side, compare input, output, context-window, and indicative speed data, and see the price gap and estimated savings per 1 million input tokens. Pricing is requested from the LiteLLM Model Catalog API, converted from per-token costs into dollar-per-million-token figures, and cached in local storage for 24 hours. Because the API does not provide the required CORS header, browsers may block live requests; the app then displays a bundled snapshot and identifies the data as cached. The static site requires no backend and is built with React 19, TypeScript, Vite 8, and Tailwind CSS 4. Its repository includes accessibility features, unit tests, linting, type checks, and pricing validation. A workflow refreshes pricing every six hours, opens a pull request when prices change, flags jumps of at least three times for review, and separately checks source links weekly. The project is released under GPLv3 or later.