wtf is athanor-lite?

Athanor Lite is a Windows desktop app that helps you figure out which free, open source AI models your computer can actually run, then gets you chatting with one in a single click. It scans your hardware, including your GPU, CPU, memory, and disk, and shows you a live dashboard with real numbers instead of guesswork. It even runs a built in speed test that measures how many tokens per second your machine can produce. The app includes a model browser with a curated list of verified AI models, each one showing exact file size and a checksum so you know the download is legitimate. For every version of every model, Athanor Lite computes a fit verdict based on your specific GPU: whether it fits fully in memory, fits tightly, needs to split between GPU and CPU, runs on CPU only, or is too big for your machine altogether. The model best suited to your hardware is highlighted at the top. Downloads can be paused and resumed, are checked against free disk space first, and are verified after they finish. Once installed, models live in a library where you can watch downloads progress, start a chat with any model in one press, or remove it later to free up disk space. If you already use Ollama, Athanor Lite can find your existing model library and adopt those files in place using hard links, so nothing is copied or re downloaded. Everything runs locally through a bundled version of llama.cpp, a tool for running AI models on your own hardware. There is no account to create, no cloud service involved, and no data sent anywhere unless you specifically choose to share diagnostic information, in which case the app shows you exactly what would be sent before you agree. To build it yourself you need Node.js, Rust, and the Protocol Buffers compiler. An NVIDIA GPU is recommended for faster performance, but the app still works on CPU alone, just more slowly. Athanor Lite is a free companion to a paid version called Athanor, which adds document search, tool connections, and multi model comparison features.