| --- |
| title: openTARS |
| emoji: 🤖 |
| colorFrom: indigo |
| colorTo: gray |
| sdk: static |
| pinned: true |
| --- |
| |
| <img src="https://huggingface.co/spaces/OpenTARS/README/resolve/main/banner.png" alt="openTARS: a local AI assistant for your computer" width="100%"> |
|
|
| # Project OpenTARS |
|
|
| **openTARS is an open source AI assistant that runs on your own computer.** Ask in your own words: it opens programs, clicks buttons by their name, types, searches the web, runs commands and looks at your screen. By default nothing leaves your machine, and there is no account or subscription. It runs on Linux, Windows and macOS. |
|
|
| ## Models |
|
|
| | Model | What it is | Size | |
| |---|---|---| |
| | [Gargantua 2](https://huggingface.co/OpenTARS/Gargantua-2) | The main model: Qwen3.5 2B fine-tuned for the assistant, with vision | 1.2 GB + 0.36 GB vision | |
| | [Gargantua 2 Lite](https://huggingface.co/OpenTARS/Gargantua-2-Lite) | A 0.8B model that runs on the CPU, for computers without a graphics card | 0.5 GB + 0.12 GB vision | |
| | [Murph](https://huggingface.co/OpenTARS/Murph) | A tiny request classifier that picks the kind of task in 0.2 ms, with no GPU | 4 MB | |
|
|
| The models are published as GGUF files with SHA-256 checksums. Every model card lists the tests we ran and the known limits. These are our own tests on one machine, not public benchmarks. |
|
|
| ## The app |
|
|
| - [openTARS on Hugging Face](https://huggingface.co/OpenTARS/openTARS): a mirror of the app and the Ubuntu/Debian package. |
| - [GitHub](https://github.com/project-opentars/OpenTARS): source code, releases and issues (MIT). |
| - [Website](https://project-opentars.github.io/OpenTARS/): downloads and install commands for Linux, Windows and macOS. |
|
|
| ## Contact |
|
|
| projectopentars@gmail.com |
|
|
| The openTARS code is MIT. The Gargantua models are Apache-2.0, based on Qwen3.5 (Apache-2.0). |
|
|
