Google has introduced an experimental app that may run an AI (AI) model directly on a cell phone without counting on the cloud.
Google quietly launched the Google AI Edge Gallery app, which allows users to run a wide range of open source AI models on mobile phones.
The app brings out the models released on the AI ​​development platform, Herging Face, and supports them without the Web. The Android version is now released first, and the iOS version can be released soon.
The AI ​​Edge Gallery allows users to look and download AI models suitable for various tasks resembling image creation, query response, code writing and editing. The app works offline using the processor within the device.
Google calls the app as ‘Experimental Alpha Release’, GitHubIt was possible to put in through.
The most important screen of the app is supplied with a shortcut icon connected to major AI functions resembling ‘AI chat’ and ‘image query’, and an inventory of AI models suitable for that task will appear once you press each function. For instance, Google’s lightweight model ‘Gemma 3N’ can be included.
It also features a function called ‘Prompt Lab’, which lets you run a single-task resembling text summary and rewriting. Users can high quality adjust the behavior of the model through various templates and settings.
Google warns that the speed and quality of execution can vary depending on the device’s performance. The more high -performance mobile phones are, the faster the model execution speed, however the larger the model size, the longer the work processing time can take longer.
Using this app, you possibly can get out of the cloud and satisfy personal data protection and offline utilization. Nonetheless, there isn’t a explanation of the aim or purpose of launch.
The AI ​​Edge Gallery may be used freely for various purposes, including the Apache 2.0 license, and business use.
By Park Chan, reporter cpark@aitimes.com
