lm-supply

.NET library for on-demand local AI model inference — zero bundled models, lazy loading, hardware-aware GPU/CPU selection, 10 task types including embeddings, generation, vision, and audio.

GitHub

Official source

Start locally with the CLI guide