Running models locally, four catalog entries: - ModelHub: a Mac menu bar app for managing local LLMs. - Swiftlet: runs large open models locally with a small memory footprint, on Mac and iPhone. - Turbo Fieldfare: an open-source inference engine for M-series Macs. - Compute:Arena: community-submitted benchmarks for running AI models locally. If you run models on your own machine, what decides which runtime stays installed?