One day, one generative AI tool

Google AI Edge Gallery (Github: https://github.com/google-ai-edge/gallery, direct IOS and Android Appstore links via the Play Store) is an experimental mobile app from Google based on the Gemma models.

After loading the chosen model from Gemma-3n-E2B-it (3.39 GB), Gemma-3n-E4B-it (4.65 GB), or Gemma-3n-1B-it (565 MB) via a Hugging Face account, they run locally and offline directly on the device.

4 modules offered: “Ask Image” for image analysis, “Audio Scribe” for the retranslation of recorded or wav audio, “Prompt Lab” with predefined instructions (summarizing, rewriting or transcribing code), “AI Chat” for a text chatbot.

Note: average performance of image analysis effective on everyday images, import of down wav file during tests and tendency to hallucinations, especially in the “Ask image” mode.

Free, unlimited and open-source.

Text tests with Art Poétique, Verlaine and personal images. (https://fr.wikipedia.org/wiki/Art_po%C3%A9tique_(Verlaine)).