Platform Support

Understanding platform requirements and compatibility for expo-ai-kit.

Overview

expo-ai-kit provides on-device AI capabilities by leveraging native platform frameworks. Both iOS and Android are fully supported.

There are two model paths. The built-in OS models — Apple Foundation Models on iOS and ML Kit on Android — are managed by the operating system rather than bundled with your app. Android may download its model on first use. Downloadable models (Gemma, Qwen, Phi via LiteRT-LM) run on both platforms and broaden support to devices without a built-in model, as long as they have enough RAM. See the Models guide.

Supported

Unsupported

iOS Support

iOS iOS 26+ provides full on-device AI support.

How It Works

expo-ai-kit uses Apple's Foundation Models framework introduced in iOS 26. The on-device language model runs entirely locally with no internet connection required.

Android Support

Android Android provides full on-device AI support via Google's ML Kit.

How It Works

expo-ai-kit uses Google's ML Kit. The model may need to be downloaded on first use on supported devices. Check the supported devices list for compatibility.

For Android-specific setup instructions, see the Android Setup Guide.

Feature Comparison

Feature availability by platform:

Graceful Degradation

Design your app to handle cases where on-device AI isn't available. Always check availability at runtime:

import { useState, useEffect } from 'react';
import {
  isAvailable,
  prepareBuiltInModel,
  sendMessage,
  type LLMMessage,
} from 'expo-ai-kit';

export function useAI() {
  const [available, setAvailable] = useState(false);

  useEffect(() => {
    void (async () => {
      if (!(await isAvailable())) return;
      await prepareBuiltInModel();
      setAvailable(true);
    })();
  }, []);

  const askAI = async (question: string) => {
    if (!available) {
      // Use cloud AI fallback or show appropriate message
      return null;
    }

    const response = await sendMessage([
      { role: 'user', content: question }
    ]);
    return response.text;
  };

  return { available, askAI };
}