Web giant Google is solidifying its standing in the AI race with the launch of its latest model, Gemini 3. The move closely follows similar model upgrades from major players like Anthropic and OpenAI and is meant to improve upon prior iterations in terms of multimodal understanding and agentic and vibe software coding.
Gemini’s core capabilities include synthesizing information, web development and complex project planning. Gemini 3 is built to provide concise and direct responses to offer improved insights and add nuance to interactions. It also features a Deep Think mode that flexes its multimodal understanding capabilities. The company launched Gemini 3 Pro in preview, providing access to it across various Google products.
The Gemini app is newly equipped with the latest model, enabling new features such as a redesigned look and a more organized layout. Gemini is also expanding to produce generative interfaces. These immersive, magazine-style views featuring photos and modules will be created instantly in response to user prompts. For example, after requesting Gemini to plan a three-day trip to Rome, the generated interface would provide a visual itinerary and invite further input to help tailor its results.
In addition, the experimental Gemini Agent feature will be rolled out, beginning with Google AI Ultra members, to support complex, multi-step tasks directly in Gemini.
