Last updated: 11/16/2026Last verified: 2026-11-11
Google AI Edge Gallery is Google’s open-source app for running open-weight models on your own phone, tablet or Mac: chat with on-device ASR and TTS, camera-based image Q&A, media search, prompt lab and model benchmarking, with no cloud roundtrip. The November 2026 update added on-device speech recognition and speech synthesis plus the XR Language Teacher showcase.
What is Google AI Edge Gallery?
Google AI Edge Gallery is a free, open-source app from Google for running open-weight AI models on your own device. The Android build installs from Google Play, the iOS build from the App Store (that listing is iPhone-only), and macOS gets a separate DMG build. The app bundles local chat with on-device speech recognition and text-to-speech, camera-based image questions, media search, a prompt lab and model benchmarking, and it runs inference on the device instead of in the cloud. The 10 November 2026 update added the speech stack and the XR Language Teacher showcase.
Key Features
- AI Chat with on-device speech recognition and text-to-speech, plus Thinking Mode on supported models starting with the Gemma 4 family
- XR Language Teacher: point the camera at an object, ask by image, voice or text, hear the answer read aloud in the target language, and save key words as vocabulary cards; it also works hands-free on supported Android XR glasses
- Instant Media Search for natural-language search over local photos, with image-to-image similarity matching
- Video Moment Finder for locating moments in videos by description, with timeline highlighting and clip trimming
- Ask Image for camera- or gallery-based visual questions
- Audio Scribe for transcribing and translating voice recordings on device
- Prompt Lab for testing prompts with controls such as temperature and top-k
- Agent Skills, plus the experimental Mobile Actions and Tiny Garden demos, both built on a FunctionGemma 270M fine-tune
- Model management for downloading listed models or loading your own
- Model benchmarking to see how a model performs on your specific hardware
Best For
Pricing
Google AI Edge Gallery is free. The repository is Apache-2.0 licensed, Google Play lists the app as free with 1M+ downloads under the publisher Research at Google, and and the App Store lists it as Free. There is no paid tier to buy; the real costs are your own hardware and the storage used by downloaded models.
Pros & Cons
Pros
- Runs open-weight models locally on Android, iPhone and Mac hardware
- Combines chat, voice, image, video and audio features in one app
- Includes model management and benchmarking so you can compare models on your own hardware
- Ships practical demos such as XR Language Teacher and Instant Media Search
- The Apache-2.0 source code is public, so the implementation can be inspected and rebuilt
- Uses LiteRT for execution and Hugging Face for model discovery and download
Cons
- Available capabilities depend on device hardware, operating system and model compatibility
- Local models consume device storage and compute; the iOS app is 132.2 MB before any model download
- Requirements are Android 12+ and iOS 17+
- Not every model supports Thinking Mode, voice features or every app experience
- Mobile Actions and Tiny Garden are experimental, and the README calls the whole app an experimental Beta release
- Scope of verification: based on the repository README, the Google Play and App Store listings and Google's 10 November 2026 blog post; the app was not installed and no model was run
Alternatives
Desktop app for running and training local models on macOS, Windows and Linux, a heavier workflow than an on-device demo gallery.
Local speech-to-text library built on ggml, for building your own transcription pipeline instead of using a ready-made app.
Terminal tool that scores models against your RAM, CPU and GPU so you know what fits before downloading anything.
FAQ
Details
Platform
Features
- AI Chat with on-device speech recognition and text-to-speech, plus a Thinking Mode on supported models starting with the Gemma 4 family
- XR Language Teacher: point the camera at an object, ask by image, voice or text, and get an answer read aloud with saveable vocabulary cards
- Instant Media Search, Video Moment Finder, Ask Image, Audio Scribe, Prompt Lab, Agent Skills and the experimental Mobile Actions and Tiny Garden demos
- Model management and benchmarking: download listed models or load your own, then measure how each runs on your hardware (LiteRT, Hugging Face integration)
Languages
Known limitations
- Free and open source: the repository is Apache-2.0 licensed, the App Store listing shows the app at $0 with no in-app purchases, and Google Play lists it free with 1M+ downloads under the publisher “Research at Google”. There is no paid tier to buy; the cost is your own hardware and storage for downloaded models
- The repository README calls it “an experimental Beta release” and the Play listing says the app is in active development with performance depending on the device’s CPU/GPU. Requirements are Android 12+ and iOS 17+; the App Store entry is iPhone-only, 132.2 MB, and the macOS build is distributed as a 0.1.0 DMG rather than through the App Store
- The 10 November 2026 Google Developers Blog post describes the speech stack as ML Kit system models on Android (100+ languages, ~100 ms to first speech, 24 kHz audio) and open-weight .litertlm bundles elsewhere (Parakeet-TDT 0.6B, PocketTTS 100M, Kokoro-82M, Phonon 31M). The XR Language Teacher demo pairs on-device ASR with Gemini Nano on Android devices or Gemma 4 on other platforms, so quality tracks the model your device can hold
- The App Store lists English plus 94 more languages for the app shell, but this entry keeps languages conservative because model and feature coverage per language varies. Model downloads come from the app’s own list or your own files, and the developers.google.com/edge/gallery page embeds its content in a frame that returned HTTP 502 during verification, so the details here come from the repository README, the Play and App Store listings and the November 2026 blog post. The app was not installed and no model was run







