Orinth
GitHubDownload
Download

Get Orinth.

No account, no sign-up, no license key. Download the app, or clone the repo and run make dev.

Orinth Desktop

macOS — universal

One .dmg for Apple silicon and Intel. It wraps the same FastAPI and Next.js servers the source checkout runs, so there is no second implementation to keep in sync.

Apple silicon + IntelmacOS 12 or later~2.7 GB installed after first launchApache 2.0
Download the .dmgBuild from source

Windows and Linux builds are not packaged yet. Both platforms run today from source — see the installation guide.

First run, start to finish.

  1. Open the .dmg and drag Orinth to Applications

    One universal build covers both Apple silicon and Intel Macs. There is no installer to step through.

  2. Approve the first launch

    The build is ad-hoc signed but not notarized, so macOS asks once on a machine that did not build it. Right-click the app, choose Open, and confirm.

  3. Let the first launch install the ML environment

    The .dmg ships source rather than bundled runtimes, so first launch downloads and installs roughly 2.7 GB — 10 to 20 minutes, with per-step progress on screen, and it needs a network connection. Later launches start in seconds, offline.

  4. Create a project

    A first-run walkthrough opens on the Projects page. Point a project at a folder of images or text and the workspace carries you from there.

From source

Two commands to a running stack.

Requires Python 3.11 with uv, Node with pnpm, and make. The backend serves on :8000 with API docs at /docs; the app on :3000.

Full installation guide
git clone https://github.com/L007/orinth.git
cd orinth

cp .env.example .env
cp frontend/.env.example frontend/.env.local

make doctor   # verify prerequisites
make dev      # backend :8000 + frontend :3000

Where your data goes

Nowhere you didn’t send it.

Datasets, weights, predictions, and logs are written to local storage. The desktop app keeps them under ~/Library/Application Support/. Two things reach the network, both on purpose: Hugging Face Hub imports, and the chat surface’s web search toggle while it is on.

Orinth is a research and engineering instrument for measuring model behavior. It produces model outputs for human review — it does not provide clinical diagnosis or autonomous decision-making.