Odysseus and Ollama: Windows setup tutorial

This beginner tutorial walks through installing Odysseus on Windows and connecting Ollama to run models locally. The description names Windows 11 and 10; the demonstrated setup uses Windows 11. The presenter follows the repository's native Windows instructions, installs Git and Python, then uses PowerShell to clone the project and install its requirements.

One installation error comes from a command targeting Python 3.11 while the presenter has installed 3.14.5. Changing the command to target 3.14 works in this demonstration. That result does not establish compatibility for every installation. The walkthrough then covers account setup and opening the browser interface at 127.0.0.1:7000. A desktop .bat file provides a shortcut to launch the server again.

For the local LLM setup, the presenter installs Ollama, restarts Odysseus and adds Ollama under the local provider settings. A separate Command Prompt runs an ollama pull command for a 0.8 billion parameter model, described as roughly 1 GB. After refreshing Odysseus, the presenter selects the downloaded model and receives a chat response.

The presenter says chats persist after restarting the PC and describes local execution as private. The tutorial does not verify the broader claim that local execution eliminates all data-leak risks. Cookbook model downloads fail in this setup, so the demonstrated download route uses Ollama.