Google announced a number of high-profile AI innovations at the I/O 2024 conference

Google's latest I/O 2024 event showcases significant innovations in artificial intelligence that deserve a special review.

- New additions to the Gemma family of open source models based on Gemini technologies include CodeGemma for programming support, Recurrent Gemma for efficient memory utilization, and PaliGemma for handling tasks linking language and images. An update to Gemma 2 with an impressive 27 billion parameters, as opposed to the current versions with 2 and 7 billion, was also unveiled.

- The updated Gemini 1.5 Pro will be able to handle up to 2 million tokens and will be integrated into all Google products, including interfacing with Google Drive, mail service and office applications.

- Gemini Nano, the most "lightweight" model from Google, will be implemented in the Chrome desktop application and Android mobile platform. Thanks to its multimodality and autonomy, the model is able to work with different types of data even in the absence of a network.

- Veo is an innovative model capable of creating 1080p quality video clips of about a minute in length based on a text query. Veo can mimic a variety of visual and cinematic styles, as well as edit already created footage. Those interested can sign up for a waiting list.-🖥 Veo is an innovative model that can create 1080p quality video clips lasting about a minute based on a text query. Veo can mimic a variety of visual and cinematic styles, as well as edit already created footage. Those interested can sign up for a waiting list.

- Imagen 3 is an updated version of the model for image generation, offering improved understanding of text queries and the ability to create more original and detailed images.

- Project Astra is an AI assistant that will promptly answer users' questions about events captured in real time via smartphone or smart glasses.

- Music AI Sandbox is an AI-based tool for working with music compositions.