Open Source · Linux

Your computer. Your data.
Your rules.

A sovereign Linux OS with local AI models. No cloud. No subscriptions. No surveillance.

# Install DirghaOS
$ dirgha --install os
 
# Run a local AI model
$ dirgha model run llama-3.3
✓ Model loaded (7B, Q4) — 18 tok/s on CPU
 
# Launch the full workspace
$ dirgha launch
✓ Writer, Research, Studio, Build — ready

What It Runs

Everything you need. Nothing you don't control.

Local AI Models

Llama, Phi, Mistral — run on your hardware. Your conversations stay on your machine. No API keys. No data leaving your network.

Full App Suite

Writer, Research, Studio, Build — the complete Dirgha workspace, pre-installed. One OS, every tool a creator or researcher needs.

Any Open Source App

Run anything from dirgha.org's app catalog. Or install your own. Flatpak, Snap, AppImage — your machine, your choice.

Built for inference

Debian

Base system

Ollama

Model runtime

CUDA + ROCm

GPU acceleration

GNOME

Desktop environment

The Full-Stack Thesis

Why we're building an operating system

Every major computing platform was built by a company that controlled the full stack — hardware, OS, and applications. Apple did it. We're doing it for the AI era.

DirghaOS is the first step. A Linux distribution designed for AI-first computing. Pre-configured with local models, optimized for inference, and integrated with Dirgha's full application suite.

The long-term goal: a complete sovereign computer company. Not a services firm. Not a SaaS wrapper. An actual computer — designed, manufactured, and shipped worldwide.

We start with software because software is where the leverage is. An OS that makes local AI effortless. Applications that don't require a cloud account. A workspace that respects the person using it.

Then we build the hardware. Then we build the factory. Then we open both to others.

Ready when you are.

DirghaOS is in active development. The first public release targets Q3 2026.

Download DirghaOS — free Coming Q3 2026 Run on Dirgha hardware