Unsloth Docker image supports training and running 500-plus models locally

Unsloth announced on September 17 that its Docker image can train and run 500-plus models locally. The company says users can choose its new graphical interface or notebooks, with no setup required, on NVIDIA and AMD.

The update follows Unsloth’s August 11 announcement of Desktop, an open-source app for running and training models locally. Unsloth says Desktop runs on Mac, Windows and Linux and can connect Claude Code and Codex to local language models.