What is this?
Hob is a free, open-source desktop app that lets you use open-weight AI models like Gemma and gpt-oss locally on Mac, Windows and Linux, as an alternative to Claude and ChatGPT. It is private and works offline.
Hob is a free, open-source desktop app that lets you use open-weight AI models like Gemma and gpt-oss locally on Mac, Windows and Linux, as an alternative to Claude and ChatGPT. It is private and works offline.Is it a chat app or something else?It is a desktop chat-style app built with Electron, and it can also handle document work like filling out PDFs. Hob picks the best open-weight model that fits your computer and sets it up for you.
It is a desktop chat-style app built with Electron, and it can also handle document work like filling out PDFs. Hob picks the best open-weight model that fits your computer and sets it up for you.How do I get it running?Download the installer for your platform from the project's download page. Hob then picks and configures a model that fits your computer automatically.
Download the installer for your platform from the project's download page. Hob then picks and configures a model that fits your computer automatically.Does my data leave my computer?The README says it is private and works offline, which suggests models run locally on your machine. Details on data handling are not stated in the README.
The README says it is private and works offline, which suggests models run locally on your machine. Details on data handling are not stated in the README.Which models does it use?Hob runs open-weight models such as Gemma and gpt-oss, and picks the strongest one that fits your hardware.
Hob runs open-weight models such as Gemma and gpt-oss, and picks the strongest one that fits your hardware.Can it work with my own files and documents?Yes, the README highlights working through documents and filling out PDFs locally on your machine.
Yes, the README highlights working through documents and filling out PDFs locally on your machine.Is it built on llama.cpp?The project's topics list llama-cpp, which suggests it uses llama.cpp to run the local models.
The project's topics list llama-cpp, which suggests it uses llama.cpp to run the local models.What are similar projects?VideoHighlighter, yovoice, Portable-Local-Studio, baiagent-director, textgen, openwhispr, atomic-agent, ollama
VideoHighlighter, yovoice, Portable-Local-Studio, baiagent-director, textgen, openwhispr, atomic-agent, ollama