# LMSupply

A .NET library for on-demand local AI inference that ships with no bundled models, downloading and caching embeddings, generation, vision, OCR, translation, transcription, speech-synthesis and image-generation models the first time they are used.

LMSupply is a local model runner for .NET built around a minimal footprint, lazy loading and zero boilerplate: the base package ships with no bundled models, and models, tokenizers and runtime components download only on first use and are cached for reuse, with hardware-aware selection choosing GPU or CPU automatically.

Separate NuGet packages cover text embeddings and semantic reranking, text generation and chat over ONNX or GGUF, image captioning, document OCR, object detection, image segmentation, neural machine translation, Whisper speech-to-text, Piper text-to-speech, shared llama-server management for GGUF models, an optional ONNX Runtime GenAI backend, and text-to-image generation with latent consistency models. Shared core packages handle Hugging Face downloads, caching, GPU execution providers, tokenization and image preprocessing.

Built with C# for .NET, distributed through NuGet. MIT licensed.

Website: https://github.com/iyulab/lm-supply