Linux install

Install Odysseus on Linux with Docker Compose

Linux is the cleanest Odysseus host if you already run Docker. The key is to verify Docker first, keep data persistent, and capture the first-run admin password from logs.

Requirements

A current Linux distribution with Docker Engine and Docker Compose installed.
Enough RAM and disk space for Odysseus plus model files.
Optional NVIDIA GPU with working drivers if local inference needs acceleration.
Ollama, vLLM, llama.cpp, OpenRouter, or another model endpoint ready to connect.

1. Clone the official repository

Start from the upstream project and review the current README before running anything. Odysseus is moving quickly, so treat the repository as the source of truth for exact compose files and environment variables.

git clone https://github.com/pewdiepie-archdaemon/odysseus.git
cd odysseus
ls

2. Start the stack

If the repository includes a Docker Compose setup, bring it up from the project root. Use detached mode only after the first run is understood.

docker compose up

After the first successful run, you can use detached mode for normal operation.

docker compose up -d
docker compose ps

3. Save the admin password

Odysseus can print setup information during first run. If you miss it, inspect compose logs before deleting volumes or resetting state.

docker compose logs odysseus

4. Open the local web UI

The common Docker web UI URL is:

http://localhost:7000

If it does not load, go to the localhost:7000 troubleshooting guide.

5. Connect a model provider

For Ollama running on the same Linux host outside the container, the OpenAI-compatible base URL often needs Docker's host gateway name:

http://host.docker.internal:11434/v1

Read the full Ollama setup guide or the model provider guide for vLLM, llama.cpp, and OpenRouter options.

Security note: Keep Odysseus local until authentication, TLS, backups, and MCP permissions are reviewed.