Google has recently unveiled Gemini 3, a line of models touted as the company’s most powerful to date, offering both agentic and vibe-coding capabilities. The top-tier model in this series is the first to be accessible simultaneously via Google’s Gemini app as well as in Search and AI Mode.
The release of Google’s new flagship AI model, Gemini 3 Pro, is a significant event due to its availability on the Gemini app. Additionally, Google subscribers have the opportunity to use this model in AI Mode in Search, with a broader roll-out to all US users expected in the near future. In the ensuing weeks, AI Mode and AI Overviews will begin to “intelligently route your most challenging questions” to this new model while reserving simpler tasks for older models.
Gemini 3 Pro is described by Google as a “natively multimodal” model that can simultaneously process text, images, video, audio, and code, rather than tackling these elements individually. This allows users to amalgamate various information sources and ask the model to synthesize them. For example, the model can generate code for interactive learning materials like flashcards after processing a mixture of research papers, video lectures, and tutorials.
Hailing Gemini 3 as their most intelligent model yet, Google believes this latest offering will bring users’ ideas to life. The model’s enhanced understanding of context and intent behind requests aims to achieve more results with less prompting.
The new features of Gemini 3 usher in significant improvements to Google’s apps. Users, for instance, can now create more comprehensive programs within the Canvas workspace of the Gemini app. The model has the capability to recognize handwritten recipes in various languages, which can then be compiled into a multilingual family cookbook.
Google refers to the new capabilities of Gemini 3 as being able to provide ‘generative user interface experiences.’ When searching for information in AI Mode, Gemini 3 Pro will visually layout the responses to your queries. These visualizations include interactive tools and simulations that are instantaneously generated based on the user’s query.
In the following weeks, Google will also debut a more research-focused version of the model, Gemini 3 Deep Think. This upgrade is expected to outperform Gemini 3 Pro, which is already setting the pace in independent benchmarks.
Along with the new model, Google has also launched a Gemini-powered coding interface named Google Antigravity. This app is akin to other agentic IDEs, merging a prompt window with a command-line interface and a browser window. Available in a free preview, the app is compatible with macOS, Windows, and Linux.
Despite Gemini 3’s impressive features, there are concerns that Google is overly ambitious in its push for AI among users. The prominent positioning of AI Overview at the top of the Search page as well as AI Mode being the first option before all results makes it impossible to miss, potentially irking users who aren’t as enthusiastic about AI.
What is the Gemini 3 Pro?
Gemini 3 Pro is Google’s new flagship AI model, capable of processing different types of data simultaneously, including text, images, video, audio, and code.
What are the new features introduced in Gemini 3?
Gemini 3 introduces ‘generative user interface experiences.’ It also allows users to create more comprehensive programs within the Canvas workspace of the Gemini app, such as recognising handwritten recipes in various languages and compiling them into a multilingual family cookbook.
What is Google Antigravity?
Google Antigravity is a Gemini-powered coding interface that combines a prompt window with a command-line interface and a browser window. The app is compatible with macOS, Windows, and Linux.
