Felix Kjellberg, the YouTube creator known as “PewDiePie,” released Odysseus on May 31, a free, open-source, self-hosted AI workspace that functions as a local-first alternative to subscription AI platforms, including ChatGPT and Claude. The project collected approximately 20,000 GitHub stars within 24 hours of launch and reached roughly 66,000 stars and 8,100 forks as of June 10.

The GitHub repository describes Odysseus as “the self-hosted version of the UI experience you get from ChatGPT and Claude,” with support for local models and external APIs from providers including OpenAI, Anthropic, Google, and DeepSeek. The official landing page describes the project as “local-first, privacy-first, and no telemetry. Just you and your models.”

Kjellberg, who has 110 million YouTube subscribers, disclosed in the launch video that most of the codebase was written using AI models. “I thought it was funny as a meme,” he said. “I’m like, ‘Look, the computer is building the computer.'” Version 1.0 carries an MIT license.

Feature Set

Odysseus packages multiple AI tools into a single workspace. Core features include:

• Multi-turn chat interface
• Autonomous agents with shell access and file editing
• Deep research mode that synthesizes web sources into structured reports
• Persistent memory using ChromaDB and fastembed
• Email assistant over IMAP and SMTP
• Document editor with markdown and HTML support
• Notes and task management
• Calendar integration with CalDAV support across services, including Nextcloud, Apple Calendar, and Fastmail
• Image editor with background removal
• Model comparison tool that sends a single prompt to multiple models simultaneously and returns results side by side

A module called “Cookbook” scans user hardware and recommends compatible models from a catalogue of more than 270 options, then serves them directly into the workspace. Supported local runtimes include Ollama, llama.cpp, and vLLM. Odysseus supports any Hugging Face model, as well as GitHub Copilot.

The platform restricts shell and file access to admin-level users by default. The GitHub documentation specifies that MCP server management, API tokens, and model serving are also admin-gated. Agents use Playwright for web browsing, enabling page navigation, screenshots, and vision capabilities.

Data Ownership and Privacy

Kjellberg framed the release explicitly around data control. “The more you share about yourself with AI, the better it becomes,” he said in the launch video. “But the more you do that, the more you’re handing over a huge piece of yourself to all these giant tech companies.”

All user data in Odysseus, including sessions, messages, documents, memory, and settings, resides in a local directory on the user’s own machine. The platform does not transmit data to external servers unless the user opts into a cloud API. The deep research module was adapted from work by Tongyi Labs; the autonomous agent system builds on the opencode project.

The GitHub documentation notes that Odysseus carries a security profile comparable to an admin console, given its capabilities around shell access, file uploads, email and calendar integrations, and API token management. It advises keeping authentication enabled and not exposing the application to the public internet without HTTPS and a trusted reverse proxy.

Community Response and Open-Source Model

The repository reached approximately 66,000 stars and 8,100 forks within ten days of launch, following a distribution path that moved from Kjellberg’s YouTube audience to GitHub Trending, then to AI-focused accounts on X, self-hosting communities on Reddit, and Hacker News. By launch week, the project had 88 contributors and more than 860 open issues and pull requests. Kjellberg posted an open call for maintainers in the project’s GitHub Discussions.

PewDiePie just embarrassed every AI startup in Silicon Valley.

He built a better local AI workspace than most funded companies. Gave it away for free. And hit 20,000 GitHub stars before most people woke up.

The project is called Odysseus. And the story behind it is more… pic.twitter.com/kFfftBsDM0

— Elias Al (@iam_elias1) June 3, 2026

Source: @iam_elias1

Kjellberg stated the project will remain free permanently. “This project will never cost any money,” he said in the launch video. “The war on big tech has just begun.”

The GitHub documentation includes setup instructions for Docker, Linux, macOS, and Windows. Kjellberg said in the launch video that contributors would need to build platform-specific ports independently.

Subscribe to Our Newsletter

Check Out Our Podcast