Setup guide

Install Vane with Docker

The Vane project recommends Docker for a first installation. Its all-in-one image includes SearXNG, so you can start with one container and configure a model in the browser.

Documentation-based walkthrough, checked October 22, 2026. The command below comes from the official README; this guide does not claim an independent installation test.

Before you begin

Have Docker running on the computer or server where you want Vane to live. Docker Desktop is the usual Docker route on Windows and macOS; Linux hosts can use Docker Engine. Make sure port 3000 is available. Decide whether you will connect a local model through Ollama or a supported cloud model provider. If you choose a cloud provider, keep its API key in Vane's setup screen and out of public pages.

The single-container command below is the same across those systems. Connecting a model running on the Docker host is where operating-system-specific addresses matter; see the Ollama page.

1. Run the official image

The project's README currently gives this command for the bundled image:

docker run -d -p 3000:3000 -v vane-data:/home/vane/data --name vane itzcrazykns1337/vane:latest

The published command maps the service to port 3000 and mounts a named volume for persistent data. For a different port or storage layout, consult the official installation section before changing it.

2. Open the setup screen

On the machine running Docker, visit http://localhost:3000. The official README says the first-run screen lets you choose models and configure provider details. If the page does not open, check that Docker started the container and that port 3000 is not occupied by another service.

3. Connect a model

Choose Ollama if you want model inference on your own machine, or configure one of the cloud providers supported by your Vane version. For Ollama across the Docker boundary, see our connection guide and the project's troubleshooting notes.

4. Verify an answer

Ask a question whose answer you can check in a primary source. Confirm that Vane returns an answer, shows source links, and lets you open those links. A reachable home screen alone does not prove search and generation are working together.

Keep this distinction in mind

Vane is a self-hosted application. This website is only a guide and does not run or provide a hosted Vane instance.

If setup fails

Run docker ps -a --filter name=vane to see whether the container is running. Read docker logs vane for startup errors. If port 3000 is occupied, stop the conflicting service or review the host-side port mapping in the official installation guide. If the page loads but answers fail, check the selected model and the Ollama connection separately from SearXNG search.

Updates, data and exposure

The named vane-data volume keeps application data beyond a container restart. Before updating, record the current image and back up any data you need. Follow the official update instructions for the current image. The command above exposes port 3000 on the host; do not expose it to the public internet without reviewing access controls and the project's security guidance.

Next step

If you want a local model, continue with Connect Vane to Ollama. If you are still choosing a tool, read Vane vs Perplexity or the alternatives guide.

Source

Vane README: Installation, checked October 22, 2026. Software changes; verify current commands at the source before running them.