See exactly what AI your machine can run — then run it. Free, local, private.
Athanor Lite is a Windows desktop app that scans your hardware, tells you which open-source AI models actually fit your GPU (with honest memory math, not guesswork), and gets you from "which model?" to a private, streaming chat in one click. Everything runs on your machine. Nothing you type ever leaves it.
Built by Black Box Analytics.
Hardware dashboard — a live instrument cluster for your machine: GPU identification (architecture, driver, CUDA), VRAM power ring, CPU / memory / GPU / disk meters at 1 Hz, and a built-in speed benchmark that measures real tokens-per-second on your hardware.
Model browser with honest recommendations — a curated catalog of verified models (exact file sizes, SHA-256 checksums). Every quant of every model gets a fit verdict computed from your GPU: fits fully, fits tight, needs a GPU+CPU split, CPU-only, or exceeds this machine. The best pick for your hardware leads the page. Downloads are resumable, checksum-verified, and pre-flighted against free disk space.
Library — everything installed in one place: watch downloads land with live progress, then start chatting with any model in one press, or reclaim the disk with a two-step delete.
Adopt your Ollama library — already using Ollama? Athanor Lite finds your models and adopts them in place with hard links. Zero bytes copied, zero bytes re-downloaded.
Private by construction — models run locally through a managed llama.cpp runtime. No account, no cloud, no telemetry unless you explicitly opt in (and the app shows you the literal payload before you do).
- Windows 10/11 (64-bit)
- WebView2 Runtime — ships with Windows 11; Windows 10 users can get it from Microsoft
- Ollama (optional) — only needed if you want to adopt an existing Ollama model library in place
- An NVIDIA GPU is recommended for accelerated inference; without one, models run on the CPU (slower, still private)
Toolchain:
- Node.js 20+
- Rust (stable, MSVC toolchain)
- protoc (Protocol
Buffers compiler) on your
PATH, or point thePROTOCenvironment variable atprotoc.exe— required by a build-time dependency
Build and run:
npm install
npm run tauri:lite:dev # run the Lite desktop app in development
npm run tauri:lite:build # produce a release buildUI-only development in a browser (synthetic hardware data, no Rust toolchain needed):
npm run dev:liteBackend tests:
cd src-tauri
cargo testAthanor Lite is the free edition of Athanor — purpose-built AI workspaces with document knowledge (RAG), tool connections (MCP), multi-model compare, fine-tuning, and more.
© 2026 Tony Winslow / Black Box Analytics. Source is provided for building, evaluation, and personal use — see LICENSE for terms.
Designed and built by Tony Winslow — Black Box Analytics.