Windows WSL2 install
For Windows users who want local models, Docker Desktop, WSL2, and optional NVIDIA acceleration.
Install hub
Install Odysseus on the hardware you already have.
Odysseus can be simple or painful depending on Docker, Python, GPU drivers, Ollama, and network exposure. Start with the path that matches your machine.
Before you run commands
Check the official GitHub README for the current install command. This guide explains decisions, risks, and common failure points.
macOS native install
Use the official macOS launch script, then connect native Ollama for local models.
Docker install checklist
Best for repeatable setup, servers, and homelabs where isolation matters.
GPU detection fixes
NVIDIA Container Toolkit, WSL integration, and common driver mismatches.
Ollama connection
Choose localhost or host.docker.internal depending on where Odysseus runs.
Decision tree
Pick the right setup path
I already use Docker.
Use the Docker path and verify ports, volumes, and model connectivity. I am on Windows.
Use WSL2 for a cleaner Linux-like environment and fewer dependency conflicts. I only need local models first.
Set up Ollama, pull a small model, then connect Odysseus. I installed it but it does not work.
Search by symptom: port, password, GPU, model, SearXNG, or ChromaDB.
Use the Docker path and verify ports, volumes, and model connectivity. I am on Windows.
Use WSL2 for a cleaner Linux-like environment and fewer dependency conflicts. I only need local models first.
Set up Ollama, pull a small model, then connect Odysseus. I installed it but it does not work.
Search by symptom: port, password, GPU, model, SearXNG, or ChromaDB.