Introduction

The RunAnywhere React Native SDK runs language, speech, and vision models on the device. Prompts, audio, and images stay on the phone, and everything keeps working offline.

import { RunAnywhere } from '@runanywhere/core'
import { LlamaCPP } from '@runanywhere/llamacpp'

await LlamaCPP.register()
await RunAnywhere.initialize({})

const result = await RunAnywhere.llm.generate('Name three colours')
console.log(result.text)

Requirements

Item	Minimum
React Native	0.83.1 (peer dependency)
React	19.0.0 (peer dependency)
iOS	17.5
Android	API 24, compile and target SDK 36

Run on a physical device. The Android emulator has no arm64 native libraries for these backends, and Apple MLX only executes on physical iOS hardware.

Packages

Package	Published version	Role
`@runanywhere/core`	0.20.24	SDK lifecycle and every capability API. Required.
`@runanywhere/llamacpp`	0.20.24	LlamaCPP backend: GGUF LLM and VLM
`@runanywhere/onnx`	0.20.24	ONNX and Sherpa: STT, TTS, VAD
`@runanywhere/mlx`	0.20.24	Apple MLX, physical iOS devices only
`@runanywhere/qhexrt`	0.20.19	Qualcomm Hexagon NPU, Android arm64-v8a only

@runanywhere/qhexrt publishes at 0.20.19, not 0.20.24. Four of the five move together; QHexRT does not.

The namespaces

Namespace	What it does
`llm`	Text generation, streaming, tools, structured output
`vlm`	Image plus prompt
`stt`	Speech to text
`tts`	Text to speech
`vad`	Voice activity detection
`voice`	A full spoken conversation
`rag`	Retrieval over your own documents
`models`	Registry, download, load, unload
`embeddings`	Text vectors
`rerank`	Reorder results against a query
`lora`	Adapters on the loaded base model
`images`	Image generation
`diarization`	Who spoke when
`segmentation`	Labelled regions in an image
`cua`	Computer-use prompt and parse scaffold

Plus five platform services: storage, logging, auth, pluginLoader, solutions. Streaming is AsyncIterable throughout.

Differences worth knowing up front

React Native’s surface is close to the other SDKs but not identical:

Input constructors are plural: AudioInputs and ImageInputs.
There is no stt.openStream or vad.openStream. Use the pull-shaped transcribeStream and detectStream.
There is no models.get. There is models.loaded(category).
RunAnywhere.deviceId is a promise.
capabilities() is synchronous.

Where to go next

Installation

Packages and native setup

Quick Start

A working chat

Text generation

Prompts and options

Voice agent

Speech in, speech out

Getting Started

Swift SDK

Kotlin SDK

React Native SDK

Flutter SDK

Web SDK

Electron SDK

Vibe Coding

Requirements

Packages

The namespaces

Differences worth knowing up front

Where to go next

Installation

Quick Start

Text generation

Voice agent

​Requirements

​Packages

​The namespaces

​Differences worth knowing up front

​Where to go next

Installation

Quick Start

Text generation

Voice agent

Requirements

Packages

The namespaces

Differences worth knowing up front

Where to go next