React Native ExecuTorch
An on-device AI inference library for React Native, built on Meta's ExecuTorch runtime, that runs machine learning models on a phone with no network calls.
React Native ExecuTorch lets a React Native app run machine learning models directly on a user's device, using Meta's ExecuTorch inference runtime. It ships with a curated set of pre-exported models for computer vision, language, text-to-speech and transcription, hosted on Hugging Face, and it also accepts custom .pte models.
The library offers ready-to-use hooks for common tasks with automatic caching and lifecycle management, plus a lower-level runtime for custom pipelines built from tensor operations, native operators and worklet threading. It accelerates execution through XNNPACK on CPU, Core ML and MLX on Apple Silicon, and Vulkan on Android GPUs, and requires React Native's New Architecture. It is written in TypeScript.