What it is
A set of shell and PowerShell scripts that download PrismML's ternary Bonsai 2 models (2B to 27B) and run them locally through a llama.cpp fork or MLX on Apple Silicon. It starts a local server with chat, image input, tool calling and MCP support, and can optionally install Open WebUI and a code interpreter. People who want to try heavily compressed open models on a laptop or desktop GPU use it to get running in a couple of commands.