Odysseus - An open-source, locally self-hosted AI workspace

What is Odysseus?

Odysseus is an open-source, self-hosted AI workspace that provides a chat interface similar to ChatGPT/Claude. The tools run entirely locally, ensuring data remains on the local machine and privacy is zero. The tool integrates over ten functions, including chat, agent, deep research, notes, tasks, calendar, email, and model comparison. It supports various model backends such as vLLM, Ollama, and OpenRouter, making it a private AI hub for individuals and teams.

Main functions of Odysseus

• AI chatIt supports communication between local models (vLLM, llama.cpp, Ollama) and APIs (OpenAI, OpenRouter), making it very easy to add models.
• AgentIt can complete tasks autonomously after being equipped with tools, supporting MCP, web pages, files, shells, skills, and memories.
• Cookbook Model LibraryScans hardware and recommends compatible models, allowing for one-click download and startup, supporting formats such as GGUF, FP8, and AWQ.
• In-depth researchIt executes in multiple steps, automatically collects, reads, and synthesizes sources to generate beautifully visualized reports.
• Model blind test comparisonSide-by-side comparison of multiple model outputs, completely anonymous voting, eliminating bias.
• Document EditorA multi-tag Markdown/HTML/CSV editor that uses AI to assist writing rather than dominate it.
• Memory and Skills: Based on ChromaDB + fastembed persistent vector memory, the agent evolves over time.
• Email clientBuilt-in IMAP/SMTP inbox, supporting AI intelligent classification, automatic summarization, automatic reply to drafts, and spam filtering.
• Notes and TasksFeatures include quick notes, to-do lists, and scheduled tasks, supporting ntfy, browser, and email reminders.
• Calendar ManagementLocal priority CalDAV calendar, can sync with Radicale / Nextcloud / Apple / Fastmail.
• Mobile supportResponsive PWA design, supports touch gestures, and is available on mobile devices.
• More toolsFeatures include: built-in image editor, theme editor, web search, file upload (visual + PDF), 2FA, etc.

Odysseus's technical principles

• Front-end architectureIt builds a responsive PWA interface based on modern web technology stack, supports theme editors and multi-tab document editors (Markdown / HTML / CSV), achieves a native cross-device experience through browser installation, and integrates 2FA authentication to ensure access security.
• Model Inference LayerBy connecting to multiple backends through a unified abstraction layer, the local side supports inference engines such as vLLM, llama.cpp, and Ollam, while the cloud side is compatible with APIs such as OpenAI and OpenRouter, enabling unified scheduling and seamless switching of multiple models.
• Agent Execution FrameworkBased on the OpenCode and MCP (Model Context Protocol) standards, it builds a tool calling system, giving LLM the ability to access the Shell, web pages, and file systems, and supports autonomous task planning, multi-step execution, and toolchain orchestration.
• Memory retrieval systemIt uses ChromaDB as the local vector database and combines it with fastembed (ONNX runtime) to complete the embedding and generation locally. It supports dual modes of vector similarity retrieval and keyword retrieval, and realizes persistent memory and skill import and export.
• Cookbook model adaptationIt scans the hardware VRAM capacity using llmfit, automatically assesses the compatibility score between quantization formats such as GGUF / FP8 / AWQ and the current hardware, and starts the vLLM or llama.cpp background service with one click. It relies on tmux to protect the download and inference processes.
• Deep Research EngineDrawing inspiration from the architecture of Tongyi DeepResearch, this approach enables multi-step information collection, source credibility assessment, reading and extraction, and comprehensive reasoning, ultimately generating a structured research report with visual presentation.

How to use Odysseus

• Docker deployment:
  • Clone the repository to your local machine:git clone https://github.com/pewdiepie-archdaemon/odysseus.git
  • Enter the project directory:cd odysseus
  • Copy environment configuration files:cp .env.example .env
  • Build and start the container:docker compose up -d --build
  • Browser access http://localhost:7000 Log in to the system.
  • View the terminal or Docker logs to obtain the temporary administrator password and complete the first login.
• Apple Silicon (M series) native deployment:
  • Clone the repository to your local machine:git clone https://github.com/pewdiepie-archdaemon/odysseus.git
  • Enter the project directory:cd odysseus
  • Execute startup script:./start-macos.sh
  • Browser access http://127.0.0.1:7860 It is ready to use and supports Metal GPU acceleration.
• Windows one-click deployment:
  • Clone the repository to your local machine:git clone https://github.com/pewdiepie-archdaemon/odysseus.git
  • Enter the project directory:cd odysseus
  • Execute PowerShell startup script:powershell -ExecutionPolicy Bypass -File .\launch-windows.ps1
  • After completing the automatic installation as prompted by the script, you can use it by accessing the local address in your browser.

Odysseus's core advantages

• Local priority, privacy priorityAll data is stored locally, eliminating the need to upload to the cloud and completely preventing privacy leaks and backdoor risks.
• Ready to use right out of the boxIt can run with the default configuration, without the need for complicated environment setup, and can be deployed with one click across multiple platforms including Docker, native, and macOS Apps.
• All-in-One WorkspaceIt integrates chat, research, notes, tasks, calendar, and email into a unified interface, reducing the need to switch between tools.
• Open Model EcosystemIt is not bound to a single model, is compatible with local inference and cloud API, and Cookbook automatically matches hardware and models.
• Agent can evolvePersistent memory and a skills system enable the agent to understand users better the more it is used, and support MCP standard extension tools.

Odysseus project address

• Project official websitehttps://pewdiepie-archdaemon.github.io/odysseus/
• GitHub repositoryhttps://github.com/pewdiepie-archdaemon/odysseus

Comparison of Odysseus with similar competing products

Application scenarios of Odysseus

• Personal privacy AI assistantSecurely use large models on local devices; sensitive data never leaves the local machine.
• Developer local LLM labIt enables rapid switching, comparison, and evaluation of different open-source models, and allows for blind testing and performance analysis.
• In-depth research by researchers: Utilize Agent to automatically collect literature, generate structured research reports, and present them visually.
• Team privatization knowledge baseBy combining vector memory and document editing, an internal question-and-answer and writing collaboration platform is built.
• Personal efficiency managementIntegrating notes, to-do lists, calendars, and emails, it creates an AI-driven hub for personal productivity.