shimmy

⚡ Pure-Rust WebGPU inference engine - OpenAI-API compatible, GGUF native, runs on any GPU. No Python. No llama.cpp. Single binary.