AskOpenSource by The Arthi AI Collective

hob

A free, open-source desktop app that lets anyone chat with AI models on their own computer, offline and privately, as an alternative to ChatGPT/Claude.

View frigadehq/hob on GitHub
GitHub stars
532
Forks
—
Language
TypeScript

Project details

Repository
frigadehq/hob
Licence
MIT
Primary language
TypeScript
Catalog status
Available
Catalog check
2026-11-13

Repository description

An easier way to use AI on your own computer. A free, open-source alternative to Claude and ChatGPT that runs open-weight models like Gemma and gpt-oss locally on Mac, Windows and Linux. Private, and works offline.

Documented setup

  1. Check your computer and the current desktop instructions

    Read the linked README and project help. Confirm the release supports your OS and CPU architecture. Exact RAM, GPU and model-performance fit have not been verified here; stop and clarify those requirements when they matter.

  2. Follow the documented local desktop setup

    Download the matching packaged app and follow the official desktop instructions. Initial application and model setup needs an internet connection. Use the documented local mode; do not add cloud services or terminal steps to this route.

  3. Check one document question against its source

    Follow Hob's documented document workflow with a small, non-sensitive PDF and ask a question whose answer you know. Compare the answer with the PDF, and confirm local processing before relying on it. No particular import control or model size is assumed here.

Requirements and limits

  • Scope: the desktop app's documented offline document and PDF use with local open-weight models. The README says Hob chooses a model to fit the computer; this is not a measured hardware guarantee.
  • Free use means no mandatory software or service fees for this documented local mode. It does not establish zero total cost: compatible hardware, storage, electricity, internet access, and model downloads depend on your setup.
  • Initial application and model downloads require an internet connection. Listed downloads cover Apple Silicon Mac, Windows x64, and Linux x86_64; confirm architecture before installation.
  • Exact minimum RAM, CPU-only performance, model size, PDF types, and a runnable first result have not been verified here. Do not promise 8 GB or Intel Mac compatibility.

Sources

Explore by topic

More tools in Security & privacy