A fast Rust LLM inference engine with quantization, OpenAI-compatible serving and SDKs for local and multi-GPU deployment.