Enthusiast developer felixrieseberg has brought back to life the legendary Clippy from Microsoft Office 1997, which has “grown” from a simple assistant into an AI assistant. The developer has preserved the user interface of the 1990s, paying tribute to the “great Clippy” and the visual style of Microsoft at that time.
Image source: felixrieseberg.github.io
The updated Clippy allows you to run various large language models (LLM) locally on your computer, including several modifications of Google’s Gemma 3, Microsoft’s models (Phi-4 Mini), Meta✴ (Llama 3.2), and Qwen (Qwen3). To do this, you need to download any of them in the application to your computer. To start working with the updated Clippy, just download the archive with the application and unpack it to a convenient location.
According to the developer, Clippy is easy and quick to set up thanks to the use of llama.cpp and node-llama-cpp — the application automatically selects the most efficient way to launch the AI model (Metal, CUDA, Vulkan, etc.). It also supports custom models, parameters, and hints. Clippy can also work locally in offline mode. The application checks for updates, but this feature can be disabled.
The enthusiast noted that his goal was to give users the opportunity to enjoy “a strange mix of nostalgia for 1990s technology and one of the most magical technologies that will be able to run on computers in 2025.” You can download the updated Clippy on GitHub. The application supports OS: Windows, macOS (Apple Silicon/Intel) and Linux (RPM, x64/Debian, x64).