SAN FRANCISCO, May 20 (Xinhua) -- Google I/O 2025, the company's largest annual developer conference, kicked off on Tuesday in Mountain View, the U.S. state of California, for a two-day event, featuring a series of product announcements spanning Google's entire portfolio.
One of the major highlights is Deep Think, a new "enhanced" reasoning mode for Google's flagship Gemini 2.5 Pro model. According to Google, Deep Think enables the model to evaluate multiple possible answers before responding, significantly improving its performance on select benchmarks.
Currently, Deep Think is available to trusted testers via the Gemini API, with Google stating that it is conducting additional safety evaluations before a broader release.
Another key announcement was Veo 3, the company's new Flow video editing app, which can be prompted using either text or an image. Veo 3 improves upon its predecessor, Veo 2, by generating higher-quality video content. Google claims Veo 3 is capable of adding sound effects, background noise, and even dialogue to accompany the videos it produces.
Veo 3 is available starting Tuesday through Google's Gemini chatbot app, exclusively for subscribers to the AI Ultra plan, which costs 249.99 U.S. dollars per month.
The Gemini Ultra subscription offers what Google describes as the "highest level of access" to its AI-powered apps and services.
Google also introduced Imagen 4, the latest version of its AI image generator, which is said to be faster and more powerful than Imagen 3. The company plans to release a variant of Imagen 4 that is up to 10 times faster than its predecessor.
Imagen 4 can render fine details -- such as fabrics, water droplets, and animal fur -- and supports both photorealistic and abstract styles. It can generate images in various aspect ratios and resolutions up to 2K.
Both Veo 3 and Imagen 4 will power Flow, Google's AI-driven video tool designed for filmmakers and content creators.
Google also announced that Gemini Live, which offers real-time verbal interactions, camera input, and screen sharing, will roll out this week to all iOS and Android users. Powered by Project Astra, Gemini Live enables near real-time conversations while simultaneously analyzing video from the user's smartphone camera or screen.
In the coming weeks, Gemini Live will integrate more deeply with other Google apps, including the ability to provide directions from Google Maps, create events in Google Calendar, and generate task lists using Google Tasks.
Google is rolling out AI Mode, the experimental Google Search feature that lets people ask complex, multi-part questions via an AI interface, to users in the United States this week, according to the company.
Additionally, Google unveiled new AI Workspace features for Gmail, Google Docs, and Google Vids. For example, Gmail will soon offer personalized smart replies and a new inbox-cleaning capability.
The company also previewed its new XR glasses, capable of augmented reality including live language translation. "Android XR is the first @Android platform built in the Gemini era, and it supports devices across a variety of use cases -- from headsets to glasses," Google said.
