LeapSDK Examples

This repository contains example applications demonstrating how to use the LeapSDK for on-device AI inference across multiple platforms.

Platforms

📱 iOS Examples

• LeapSloganExample: Simple SwiftUI app for slogan generation
• LeapChatExample: Comprehensive chat application with real-time streaming
• LeapAudioDemo: Audio processing and transcription demo
• LeapVLMExample: Vision language model demo for image understanding
• LeapVoiceAssistantDemo: Press-and-hold voice assistant powered by VoiceAssistantWidget from leap-ui
• RecipeGenerator: Recipe generation with constrained JSON output

Swift Package Manager integration with XcodeGen project generation.

🖥️ macOS Examples

• LeapVLMExample: Vision language model demo for image understanding
• LeapVoiceAssistantDemo: Press-and-hold voice assistant (macOS counterpart of the iOS demo)

Swift Package Manager integration with XcodeGen project generation.

🤖 Android Examples

• SloganApp: Basic slogan generator using Jetpack Compose
• LeapChat: Full-featured chat application with modern Android UI
• LeapAudioDemo: Audio input/output demo with streaming playback and interactive controls
• ShareAI: Web page summary generator
• RecipeGenerator: Recipe generation with constrained JSON output
• VLMExample: Vision language model demo for image understanding
• LeapVoiceAssistantDemo: Press-and-hold voice assistant powered by VoiceAssistantWidget from leap-ui
• LeapKoogAgent: AI agent demo integrating Koog framework with LeapSDK

Gradle-based projects using the LeapSDK Android library.

🌐 Web Examples

• LeapVoiceAssistantDemo: Kotlin/Wasm Compose-for-Web port of the voice assistant. Run with ./gradlew wasmJsBrowserDevelopmentRun (dev server) or ./gradlew wasmJsBrowserDistribution (static bundle).

Quick Start

iOS

cd iOS/LeapSloganExample
make setup && make open

Android

cd Android/SloganApp
./gradlew installDebug

Web

cd Web/LeapVoiceAssistantDemo
./gradlew wasmJsBrowserDevelopmentRun
# open http://localhost:8080

What is LeapSDK?

LeapSDK enables running AI models locally on mobile devices and in the browser using the Liquid Inference Engine. It provides:

• On-device inference - No internet required
• Real-time streaming - Token-by-token response generation
• Cross-platform - iOS, macOS, Android, and Web (Kotlin/Wasm) support
• High performance - Optimized for mobile hardware
• Easy integration - Simple API for chat and text generation

Documentation

• 📚 iOS Quick Start Guide
• 📚 Android Quick Start Guide
• 🔗 iOS SDK Repository
• 🔗 Android SDK Repository

License

See LICENSE for details.