Gemini Nano: Google's AI revolution for mobile phones, even offline

Google Announced this year twinthe most advanced artificial intelligence model to date. Android 14 Model said artificial intelligence They are combined to revolutionize tasks on mobile devices soon.

The new tool will allow advanced functionality such as creating Smart answers Grammar correction in applications without the need to connect to the network.

Gemini Nano It comes in three enhanced versions: Ultra, Pro and Nano; The latter is ready to be enabled in Settings as a developer preview in Pixel 8 Prooffering high-quality automatic responses in apps like WhatsApp.

to implement Gemini Nano in it Pixel 8 Pro It demonstrates its capabilities, enabling the application registered Summarize the content concisely, even without the Internet.

Android iCorea new system service in Android 14, makes Gemini Nano easier to access and handles model management, runtimes, and security features, all while keeping the process private and transparent.

AICore It also encourages customization through adaptation Low rank (Laura)which allows developers to create small LoRA adapters tailored to their training data, thus running a large language model tailored to specific use cases in each application.

Using new machine learning hardware, such as the latest Google Tensor TPU and NPU from companies like Qualcomm, Samsung SLSI and mediatekexpand the reach of AICore and Gemini Nano to more devices soon.

Developers are interested in exploring the possibilities it offers Gemini Nano You can sign up for the Early Access program to learn more about the tools and APIs available to build on the model on supported devices Android.

The goal is to integrate Advanced artificial intelligence research With accessible tools to foster innovations in mobile applications that improve user experience through more flexible and powerful AI.

To test the tool you must enter this Form.

This is what everyone was waiting for, the announcement Google DeepMind with Geminia flexible, multi-modal artificial intelligence (AI) model that promises to revolutionize the way technology intertwines with everyday life and business development.

Sundar Pichai, CEO of Google and Alphabet, highlighted that this new model, part of the company's initiative to be “AI first,” has cutting-edge performance on several academic benchmarks and is capable of outperforming human experts in tests. .

Gemini, which is optimized to run on devices with different capabilities (Ultra, Pro and Nano), demonstrated superior performance on 30 of 32 academic tests applied to language models, including a score of 90.0% in language comprehension. Multi-conceptual MMLU, which hides the analytical power of human specialists in fields such as mathematics, physics, history, and medicine.

Sundar Pichai He emphasized the continued acceleration of AI progress in the enterprise and the promise that these developments will be of global benefit.

The new system, the result of a broad collaborative effort at Google, is designed to efficiently understand and analyze information, including text, code, audio, image and video.