Skip to content

BBALabs/athanor-lite

Repository files navigation

Athanor Lite

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.


Features

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).

Screenshots

Prerequisites

  • 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)

Building from source

Toolchain:

  • Node.js 20+
  • Rust (stable, MSVC toolchain)
  • protoc (Protocol Buffers compiler) on your PATH, or point the PROTOC environment variable at protoc.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 build

UI-only development in a browser (synthetic hardware data, no Rust toolchain needed):

npm run dev:lite

Backend tests:

cd src-tauri
cargo test

The full version

Athanor 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.

bbasecure.com/athanor

License

© 2026 Tony Winslow / Black Box Analytics. Source is provided for building, evaluation, and personal use — see LICENSE for terms.

Credits

Designed and built by Tony WinslowBlack Box Analytics.

About

Free GPU detection and local AI model manager. Scans your hardware, recommends compatible models, and handles download + launch in one click.

Resources

License

Stars

Watchers

Forks

Packages

 
 
 

Contributors