google-ai-edge/gallery

A gallery that showcases on-device ML/GenAI use cases and allows people to try and use models locally.

What it solves

Google AI Edge Gallery is an application designed to bring high-performance, open-source Large Language Models (LLMs) and multimodal AI directly to mobile devices and macOS. It eliminates the need for cloud-based AI, allowing users to run models fully offline for increased privacy, speed, and accessibility.

How it works

The app leverages Google AI Edge APIs and the LiteRT runtime for optimized on-device execution. It integrates with Hugging Face for model discovery and downloading, allowing users to load a variety of open-source models or their own custom models.

Who it’s for

Developers and AI enthusiasts who want to experience, evaluate, and test on-device generative AI capabilities on Android, iOS, and macOS.

Highlights

  • Agent Skills: Adds tools like Wikipedia and interactive maps to LLMs to make them proactive assistants.
  • AI Chat with Thinking Mode: Supports multi-turn conversations and a reasoning process view for supported models like Gemma 4.
  • Instant Media Search & Video Moment Finder: Uses EmbeddingGemma 2 for on-device semantic search of photos and videos.
  • Ask Image: Multimodal capabilities for identifying objects and solve visual puzzles via camera or gallery.
  • Audio Scribe: Real-time on-device transcription and translation of voice recordings.
  • Prompt Lab: A workspace for testing prompts with granular control over parameters like temperature and top-k.
  • Mobile Actions & Tiny Garden: Demonstrates offline device controls and automated tasks using FunctionGemma 270m.
  • Model Management & Benchmark: Tools to download, manage, and run performance benchmarks on specific hardware.

Related

  • Project
    google-ai-edge/LiteRT-LM

    A production-ready orchestration layer for running high-performance, cross-platform LLMs on-device using hardware acceleration.

  • Dispatch
    LLM Inference on Edge: Running Local LLMs via React Native

    Hugging Face provides a comprehensive guide to building a privacy-focused mobile application using React Native and llama.rn to run quantized GGUF models locally on Android and iOS.

  • Project
    scaleapi/llm-engine

    An open-source engine for fine-tuning and serving large language models, providing tools to deploy models on hosted or self-managed Kubernetes infrastructure.

  • Project
    dineshsoudagar/local-llms-on-android

    An Android application that enables fully offline, private LLM inference on-device and can act as a local LAN server for other devices via an OpenAI-compatible API.

  • Project
    timmyy123/LLM-Hub

    An open-source mobile app for Android and iOS that enables 100% on-device LLM chat, image, video, and music generation for complete privacy.