Odysseus tutorial: install, run, and use Odysseus

What you need first

• Git, Docker, and Docker Compose if you want the most repeatable setup.
• Ollama or another model provider if you want local chat and agent responses.
• A machine with enough RAM for the model you plan to run.
• Time to read the first-run logs before closing the terminal.

1. Clone Odysseus from GitHub

The common first command users search for is:

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

Always review the current README after cloning because Odysseus is a fast-moving open-source project.

2. Run Odysseus

For Docker-based installs, start from the official compose workflow in the repository. Keep the first run in the foreground until you understand the logs.

docker compose up

After it works, run it in the background:

docker compose up -d
docker compose ps

3. Open the web UI

The local web interface is commonly opened at:

http://localhost:7000

If the page does not load, use the localhost:7000 fix guide.

4. Log in and fix credentials

Look for first-run admin credentials in terminal output or Docker logs. If the login page says invalid credentials, do not delete your data volume first. Read the invalid credentials guide.

5. Connect a model

Odysseus needs a model provider. The beginner-friendly local path is Ollama. If Odysseus runs in Docker and Ollama runs on the host, try:

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

For provider choices, compare Ollama, vLLM, llama.cpp, and OpenRouter.

6. Start using Odysseus

• Start with a normal chat to confirm model responses work.
• Add one tool or MCP server at a time.
• Test Deep Research only after search is configured.
• Keep Odysseus private until authentication and reverse proxy settings are reviewed.

For a more detailed first-workflow guide, read How to use Odysseus after installing it.

Odysseus tutorial FAQ