Google Quietly Launches Google AI Edge Gallery
At Google I/O 2025, while big announcements grabbed headlines, Google slipped out an experimental app called Google AI Edge Gallery — currently on Android, with iOS coming soon.
What it is:
• An offline AI app that runs open-source AI models (including Hugging Face’s) directly on your phone.
• No internet connection needed so your data stays on your device.
• Enables tasks like image generation, document summarization, code writing, and Q&A using local AI models like Google’s Gemma 3n.
Why it matters:
• Moves generative AI from the cloud to the device (edge computing).
• Enhances privacy, reduces dependency on connectivity, and avoids cloud data transfers.
• Empowers developers, hobbyists, and users who want AI capabilities without sending data away.
How it works:
• Simple interface with options like “AI Chat” or “Ask Image.”
• Prompt Lab lets users quickly rewrite text or get short answers.
• Performance depends on your phone’s power and model size.
Status & future:
• Still in alpha, available on GitHub under Apache 2.0 license.
• Google aims for community feedback to improve it.
• As AI models become more efficient and hardware advances, this could redefine AI usage on personal devices.