The smallest edge AI device for local LLMs

North Korean spies are running local LLMs to cause AI mischief

Homebench – Benchmark local LLMs for speed, memory, and quality

Relm – local LLMs as base-R objects, with interpretability