runanywhere-sdks

Why we included this project

Most teams building AI into a mobile app or a desktop or browser product end up wiring up llama.cpp, ONNX, and MLX by hand. RunAnywhere avoids that with one API over a shared C++ core, plus SDKs for Swift, Kotlin, Flutter, React Native, Python, and the web, so a single chat, vision, speech, RAG, embeddings, or image generation call behaves the same on every platform. It routes each request to the engine that fits the device, sending QHexRT to Snapdragon NPUs, MLX to Apple silicon, Core ML to diffusion workloads, and llama.cpp elsewhere, which removes most of the per-backend fiddling. Everything runs on-device by default, which keeps data private and makes offline use straightforward, and the structured output, tool calling, and voice-agent pipeline cover the harder integration scenarios. For a small team that wants one coherent local-AI layer rather than several separate engine bindings, it is a practical fit.