Download · v2.5.1 · MIT

Get Guaardvark.

Two commands. No signup. No telemetry. The latest release runs on Linux, with macOS and Windows support via Docker.

Two-command install

From clone to running, in under two minutes.

git clone https://github.com/guaardvark/guaardvark.git && cd guaardvark
./start.sh
# on first run: installs Python and Node deps, builds Whisper.cpp,
# runs DB migrations, builds the React frontend, starts services.
-> Web UI: http://localhost:5173
-> API: http://localhost:5000
-> Health: http://localhost:5000/api/health
Distribution channels

Pick your path.

GitHub release tarball

Versioned releases with full source. Hash verification supported.

All releases →

Git clone (recommended)

Clone the main branch, run ./start.sh, get the latest code plus easy git pull upgrades.

Repo →

PyPI CLI (guaardvark-guardian)

The CLI tool installs separately from the platform itself.

PyPI →

Docker (Linux)

For pinned, reproducible deployments. See docker-compose.yml in the repo. Runs CPU-only by default; GPU passthrough requires nvidia-docker.

Hash verification

Every GitHub release includes a SHA256SUMS file. Verify your download matches before extracting:

bash
curl -L -o guaardvark-2.5.1.tar.gz https://github.com/guaardvark/guaardvark/releases/download/v2.5.1/guaardvark-2.5.1.tar.gz
curl -L -o SHA256SUMS         https://github.com/guaardvark/guaardvark/releases/download/v2.5.1/SHA256SUMS
sha256sum -c SHA256SUMS

If the verification fails, do not run the installer. Open an issue on GitHub.

FAQ

Download questions.

Which OS does Guaardvark support?

First-class: Linux (any modern distro with Python 3.12+ and Node 20+). Supported via Docker: macOS and Windows. Native macOS/Windows installs work but aren't actively tested every release.

Do I need a GPU?

Not strictly. Most features (chat, agents, RAG) work CPU-only. Image gen, video gen, voice TTS, and large LLM inference are dramatically faster with a CUDA GPU. See pricing for hardware tiers.

How big is the download?

The Guaardvark source tarball is ~50MB. First run downloads dependencies (~2GB) and any models you choose (Ollama models are ~4–40GB depending on size). Subsequent runs reuse the cache.

Where does it install to?

Anywhere you cloned to. Guaardvark uses GUAARDVARK_ROOT env var auto-detected from the clone directory. No global system installation; you can have multiple parallel installs.

How do I upgrade?

git pull && ./start.sh. The startup script auto-runs pending migrations.

Already have a machine? Two minutes from now you have an AI lab.

Free, MIT-licensed, runs offline.