What is Gebril?
Gerbil is the easiest way to run powerful AI language models right on your own computer, no internet needed. It’s perfect if you want privacy, speed, and full control over your AI without relying on cloud services.
With a simple, clean interface, Gerbil handles all the tricky setup: downloading models, managing different backends, and using your CPU or GPU (even on older machines). Whether you’re on Windows, macOS, or Linux (including Wayland), it just works.
You can even use pre-downloaded files and go completely offline. It runs LLMs locally using KoboldCpp, so your data stays private and secure. No coding, no hassle, just point, click, and start chatting with AI.
It is great for writers, developers, researchers, or anyone who wants fast, reliable AI that respects their privacy.
The best part is it works seamlessly on Windows, Linux, and macOS (Silicon & Intel).
Features
- Run LLMs locally – Powered by KoboldCpp, a fork of llama.cpp
- Cross-platform – Native support for Windows, macOS, and Linux (including Wayland)
- Offline capable – Import pre-downloaded KoboldCpp binaries and run entirely without an internet connection
- Flexible hardware support – Works on CPU-only systems while also supporting GPU acceleration (CUDA, ROCm, Vulkan, CLBlast, Metal)
- Image generation – Built-in presets for Flux, Chroma, Qwen Image, and Z-Image workflows
- Integrated HuggingFace search – Browse models, view model cards, and download GGUF files directly from the app
- SillyTavern integration – Launch SillyTavern for advanced character interactions (requires Node.js)
- OpenWebUI integration – Launch OpenWebUI for a modern web-based chat interface (requires uv)
- Privacy-focused – Everything runs locally with no external data transmission or telemetry
License
AGPL-3.0 License.
