Gemini 3 Launches to Praise from OpenAI’s Sam Altman

Google has officially launched the Gemini 3 Pro artificial intelligence (AI) model. While currently in preview and gradually rolling out worldwide, the tech giant has begun integrating it into various products. Notably, the Gemini app and AI Mode in Search are among the first to adopt the model’s capabilities. A key new feature being introduced is “generative interfaces,” designed to provide users with a more visual and interactive information experience. The model may be so good, even competitors like OpenAI’s Sam Altman thinks its a great model (could be sarcastic or maybe not). 

Josh Woodward, Google’s Vice President of Google Labs, Gemini, and AI Studio, announced this new enhancements to the Gemini app through a blog post. While the firm uses the term “app,” these modifications will apply to both the Android and iOS apps, as well as the Gemini online interface.

The Gemini platform will be redesigned with a clean and modern aesthetic, according to the company. That means users will find it easier to start discussions, access photographs, videos, and reports they’ve generated in a new My Stuff folder, and shop for products thanks to the integration of Google’s Shopping Graph.

In a separate blog post, the company simply refers to this feature as an interface that “dynamically generates immersive visual experiences and interactive elements like web pages, games, tools, and applications with all fully customized and automatically created based on any question, instruction, or prompt.”

With two experiments, the Gemini app introduces generative interfaces for the first time. The first step is to design the layout visually. It’s a magazine-style display with photographs and modules. For instance, if a user requests a trip itinerary, Gemini might present a carousel showcasing various excursion options to choose from, a slider to set the trip duration, and a clickable bar highlighting points of interest. Based on these selections, the AI can generate an ideal plan that eliminates the need for the user to type or speak detailed instructions.

The second other option to use is to use dynamic view. The chatbot can design and write a custom user interface in real time in response to a user query, thanks to Gemini 3’s agentic coding capabilities. For example, if a user requests that Gemini “explain the Van Gogh Gallery with life context for each piece,” it will open an interactive window in which users may visually examine the content by clicking, scrolling, and sliding across sections to obtain insights.

Google is also introducing Gemini Agent, an experimental feature designed to perform multi-step tasks within the app. It can integrate with Google apps to manage calendars, set reminders, and organize inboxes by drafting email replies. Additionally, it handles web-based tasks such as booking and scheduling appointments. Currently, it’s available online for Google AI Ultra subscribers in the U.S.

The new thing about the AI Mode is that, the AI Mode can now handle more complicated searches and effectively sift through a huge number of web pages to locate contextually relevant solutions, thanks to better reasoning capacity. The new automatic model selection feature is also being added to Search, which will direct users’ more difficult questions in AI Mode and AI Overviews to Gemini 3 Pro. It will continue to employ quicker and easier models for simple enquiries.

AI Mode also sees the introduction of generative interfaces. Google claims that this will allow the AI tool to generate visual layouts for responses in real time, complete with interactive tools and simulations depending on user enquiries.

The writeup added that when the model detects that an interactive tool will help you better understand the topic, it uses its generative capabilities to code a custom simulation or tool in real-time and adds it into your response.

In an instance, the tech gaint noted that if a customer is looking for mortgage loans, Gemini 3 in AI Mode may instantly generate a personalised interactive loan calculator within the interface to assist them compare various possibilities.

When it comes to AI dominance, this year has been merciless, with every major AI company actively challenging its competitors to build better and faster LLM.

Google is joining the competition, following in the footsteps of OpenAI and Anthropic, both of which debuted their own cutting-edge models just a few months back. It’s unknown whether this combination of speed, power, and features will be sufficient. While Gemini 3 is advertised as a powerful model with strong multimodal capabilities and reasoning abilities, Google is still under pressure from both new and existing companies as the market fills with frontier models that are more capable of reasoning and increasingly appropriate for autonomous tool use.

The more important question is who will eventually dominate the AI market, and whether that leadership will come from the United States, China, or somewhere else. DeepSeek, a Chinese start-up, stunned the globe in January with its open-source strategy, demonstrating a growing wave of powerful Chinese models striving for global influence. Later, Mark Zuckerberg’s Meta created a new superintelligence unit, luring away several top OpenAI scientists, albeit no one understands exactly what the business is attempting to achieve.