Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/techjarves/Odysseus-Portable/llms.txt

Use this file to discover all available pages before exploring further.

Odysseus Portable bundles the Odysseus AI web interface and a hardware-optimized llama.cpp inference engine into a single self-contained directory. Double-click a startup script and you have a local ChatGPT-style workspace running in your browser — without touching your system PATH, installing global packages, or requiring administrator privileges.

Quickstart

Get Odysseus Portable running on Windows, macOS, or Linux in minutes

Requirements

Storage, hardware, and platform requirements before you begin

Inference Backends

Choose between llama.cpp (GGUF) and Ollama inference engines

Configuration

Environment variables, launcher config, and Hugging Face credentials

Models

Download, manage, and serve GGUF models from the web UI or models folder

Troubleshooting

Fix port conflicts, OOM errors, and USB filesystem issues

How It Works

Odysseus Portable’s orchestrator handles the entire setup automatically on every launch:
1

Hardware scan

Detects your OS, CPU architecture, RAM, and GPU (CUDA / Vulkan / Metal / CPU-only) to select the right inference binary.
2

Runtime bootstrap

Downloads and caches portable Node.js 22, Python 3.12, and a hardware-matched llama-server binary — only on first run. Subsequent launches are instant.
3

Source sync

Clones or updates the Odysseus web application from GitHub, then applies self-healing patches to ensure USB-portable compatibility.
4

Launch

Starts the inference backend and Odysseus web server, then opens http://127.0.0.1:7070 in your default browser.

Key Features

Zero System Dependencies

Node.js, Python, and llama-server are all downloaded into the project folder. Nothing is installed globally.

GPU Acceleration

Automatically selects CUDA, Vulkan, Metal, or CPU inference based on detected hardware. No manual configuration needed.

Dual Backends

Switch between llama.cpp (GGUF router with auto context scaling) and Ollama with a single environment variable.

Data Portability

All user data, models, chat history, and credentials stay inside the project folder. Unplugging leaves no trace on the host.

Cross-Platform

Runs on Windows, macOS (Intel & Apple Silicon), and Linux (x64 & ARM64) from the same directory.

USB Drive Ready

Handles exFAT/FAT32 filesystem limitations automatically with symlink-to-hardlink fallback.
Data Portability Guarantee — All databases, config files, cached runtimes, and GGUF models live inside the project folder. Unplugging your USB drive leaves no configuration traces, files, or environment variables on the host system.

Build docs developers (and LLMs) love