hob
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.
- ai-agent
- ai-assistant
- chatgpt-alternative
- claude-alternative
- desktop-app
- electron
- gemma
- gpt-oss
- linux
- llama-cpp
- llm
- local-ai
- local-llm
- macos
- mcp
- offline-ai
- open-source
- privacy
- private-ai
- windows
- Stars
- 108
- Forks
- 4
- + today
- +4
- Created
- 13d
Ranking data as of October 26, 2026 (UTC).
Star History
Today, hour by hour
Overview
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. It ranks #52 on GitTiger, gaining +4 stars on October 26, 2026 (UTC).
The project is written in TypeScript and has 4 forks. It was created 13d ago.
What it is
Hob is a desktop application from FrigadeHQ for running AI models on a personal computer instead of routing every prompt through a hosted service. The repository describes it as a free, open-source substitute for Claude and ChatGPT that works with open-weight models, naming Gemma and gpt-oss as examples. It is written in TypeScript, ships under an MIT license according to the metadata, and the description lists Mac, Windows and Linux as target platforms.
Capabilities described
Three claims come directly from the repository description. First, inference happens against models held locally rather than through a vendor API. Second, the design is private, meaning conversations stay on the machine. Third, it keeps working without a network connection. Those are the project's own characterisations and the supplied evidence contains nothing independent of them.
The repository topics add context about construction rather than features: electron, llama-cpp, local-ai, gemma and gpt-oss appear among the tags, alongside broader ai-agent and ai-assistant labels. The README preview itself carries no further detail, so there is no evidence here about the interface, model download flow, hardware requirements or memory footprint.
Getting started
No installation procedure, build steps or system requirements were supplied for this profile. The repository README is the place to look for setup instructions and any prerequisites before downloading the application.
git clone https://github.com/FrigadeHQ/hob.git
cd hob
# see README for setupFrequently Asked Questions
Is hob free to use?
The repository description calls it free and open source, and the project metadata lists an MIT license. No paid tiers, subscriptions or commercial editions are mentioned in the supplied evidence.
Which operating systems does hob run on?
The description names Mac, Windows and Linux. It does not give minimum OS versions, and no hardware requirements such as memory or disk space appear in the evidence, so the README is the place to confirm them.
Does hob need an internet connection?
The description states that it works offline, because models run locally rather than being called over the network. Whether an initial download of model weights is required is not covered by the supplied material.
Which models can hob run?
Gemma and gpt-oss are the two named in the description, and the repository topics also mention llama-cpp. The evidence does not provide a full supported-model list or say how models are obtained.