#Getting Started

@react-native-ai/core-ai lets React Native apps run app-provided Apple Core AI model bundles. Use it when you export models from apple/coreai-models and ship or download those bundles in your app.

Use @react-native-ai/apple for Apple system models. Use @react-native-ai/core-ai for custom Core AI model assets.

#Installation

npm install @react-native-ai/core-ai

To use it with the AI SDK, install ai v7 and the required polyfills:

npm install ai

#Requirements

• React Native New Architecture
• iOS 27 deployment target (with Expo, set ios.deploymentTarget to 27.0 in expo-build-properties)
• Xcode 27 or newer for building apps with Core AI
• Exported .aimodel bundles from apple/coreai-models
• A physical device. The iOS Simulator SDK has no CoreAI framework, so apps that install this package cannot be built for the simulator

#Native Setup

The pod links Apple's coreai-models Swift Package (1.0.0 up to the next major version) to itself during pod install, so there is no Xcode or config plugin step. By default it links every product the native module uses:

• CoreAILM for coreAI.languageModel(...) and sessions
• CoreAIDiffusion for coreAI.imageModel(...)
• CoreAISegmentation for CoreAI.runTask('segmentation', ...)
• CoreAIObjectDetection for CoreAI.runTask('object-detection', ...)

To link a smaller set, set CORE_AI_PRODUCTS when installing pods (with Expo, when running prebuild):

CORE_AI_PRODUCTS=CoreAILM npx pod-install
CORE_AI_PRODUCTS=CoreAILM npx expo prebuild -p ios

Calling a feature whose product is not linked throws a setup error.

Supported today: text generation, image generation, segmentation, and object detection. Embeddings, transcription, depth, super-resolution, and classification are tracked in #237.

#First iOS Model

Start with Qwen3 0.6B because Apple’s registry includes an iOS preset:

uv run coreai.llm.export Qwen/Qwen3-0.6B --platform iOS --output-dir ./models

Then load it from React Native:

import { coreAI } from '@react-native-ai/core-ai'
import { streamText } from 'ai'

const model = coreAI.languageModel({
  id: 'qwen3-0.6b',
  source: { type: 'file', uri: qwen3ModelDirectory },
  variant: 'iOS',
})

const result = streamText({
  model,
  prompt: 'Explain Core AI in one sentence.',
})

for await (const delta of result.textStream) {
  console.log(delta)
}

For a persistent native session:

await model.prepare()

const session = await model.createSession({
  instructions: 'Answer in one short paragraph.',
})

const response = await session.respond('What is on-device AI?')

await session.close()
await model.unload()

Need React or React Native expertise you can count on?

Let's talkLet's talk