In this December we get some big updates from Google. Do you know that Gemini 2.0 can make your iPhone smarter than Apple Intelligence? As, not only did we get the massive December feature drop that brought new features for Pixel devices, but we also got some major news today. Updates for Gemini 2.0 and other Google AI products like Project Astra have been announced. There is a lot which I am going to share with you.
First of all I am going to talk about Gemini 2.0. I got this announcement from an official Google blog post. The main theme of Gemini 2.0 is to make information more useful by performing tasks on your behalf. Gemini 2.0 is up to two times faster than Flash Gemini 1.5 Pro and outperforms it on key benchmarks like coding, math, and reasoning.
It supports multi model output, such as text mixed with images for conversational multi-turn editing. Multilingual audio output is also supported, which developers can customize in terms of voices, languages, and accents. And Gemini 2.0 Flash can call up tools like Google Search for factual answers and code execution.
Developers can get an experimental version today through AI Studio and Vertex AI. Gemini users can switch to a chat-optimized version starting this week, and Gemini 2.0 Flash is being tested to provide better answers on complex topics.
Next, let’s talk about Project Astra. It is Google’s attempt to create a futuristic and intuitive AI companion that can integrate seamlessly into everyday life. The last update came at Google I/O in May 2024, when Google demoed the product in smart glasses. There have been many updates since then.
Astra is still in early development, but it is built on Gemini 2.0. It can talk in multiple languages โโand can also understand accents and uncommon words. Astra can respond to prompts using Google Search, Google Lens, and Google Maps. It has an in-session memory of up to 10 minutes and can understand language at the speed of human conversation. Astra has an Android app for testing purposes, and when it goes live for consumers, it will be available through the Gemini live app.
Now we know about Project Mariner. Project Mariner is an agent that can browse or navigate the web for you. It can type, scroll, and click to complete tasks that you specify. It can understand information on the web such as text, code, images, and forms. According to the Web Voyager Benchmark, Mariner achieved a state-of-the-art result of 83.5% as a single agent.
Google has also mentioned some safety features like requiring confirmation for sensitive actions and only taking actions in the active browser tab. It is designed to prevent fraud schemes, phishing attempts, and harmful instructions. Currently, Project Mariner exists as an experimental Chrome extension, but discussions about its future are ongoing within its web ecosystem.
There are a few more announcements I’d like to mention. Gemini Advanced users will now have access to the Deep Research feature, which can search, find, and organize information for specific research questions. Deep Research can repeat the discovery process for accuracy and create a report that can be exported to Google Docs.
The second item is Jewels, an AI-powered coding agent that integrates into GitHub workflows. Jewels can work on issues, develop plans, and execute them under developer supervision.
Lastly, there is a prototype of Gemini 2.0 for games, which is a virtual gaming companion that can see your screen and answer questions about the game.
This is everything you should know about the Gemini 2.0 announcement. The main goal here is to make AI more useful. Seeing how Gemini integrates with other tools will get me excited about whatโs next. Share your thoughts in the comment section and always keep an eyes on TECHNO MINDA for tech-related latest updates.
