An easy way to try local AI with braindead-local-intelligence dataswamp.org

Solène introduces braindead-local-intelligence, a project that uses rootless Podman containers to run local language and image models, web search and an Open WebUI interface on Linux. It loads one model at a time to conserve GPU memory and targets AMD graphics cards; the default models require about 8 GB of storage, with the smallest needing roughly 4 GB of VRAM and 8 GB of RAM.