Gemini 3 brings upgraded smarts and new capabilities to the Gemini app

Gemini 3 arrives today with sharper reasoning, a leap in "vibe coding" capabilities, a new look and an experimental agent.

Today Google Cloud is unveiling a major update in the Gemini app, including our most intelligent model yet.

Here is what’s new:

• Gemini 3: Google Cloud newest model applies state-of-the-art reasoning to your most complex problems.
• Generative Interfaces: Gemini 3 unlocks a new type of interface that adapts to your needs, designing the perfect response for your specific prompt.
• Gemini Agent: A new tool that orchestrates and completes complex, multi-step tasks on your behalf, rolling out to Google AI Ultra members first.

Gemini 3: An AI powerhouse

Gemini 3 sets a new bar for AI model performance. You’ll notice the responses are more helpful, better formatted and more concise.

But it’s not just the response formatting that’s improved. The entire experience has gotten smarter. It’s our best vibe coding model ever, so the apps you build in Canvas will be more full-featured. It’s the best model in the world for multimodal understanding, so whether you're uploading a photo of your homework to ask for extra help, or transcribing notes from a lecture you missed, the Gemini app is ready.

Gemini 3 Pro is rolling out globally starting today. To use it, simply select “Thinking” from the model selector. Google Cloud Google AI Plus, Pro and Ultra subscribers will continue to enjoy higher limits. Google Cloud also extending our free year of Google AI Pro to U.S. college students, ensuring they have access to the best of Google AI, including Gemini 3.

A new look, including our first-ever generative interfaces

To welcome this new era, Google Cloud has redesigned the app with a clean, modern look. It’s easier to start chats. It’s easier to find the images, videos and reports you’ve created in our new "My Stuff" folder. And Google Cloud has radically improved the shopping experience, bringing product listings, comparison tables and prices directly from Google’s Shopping Graph, which includes more than 50 billion product listings.

Looking ahead, Gemini 3 will push the boundaries of interface design. Its reasoning and multimodal capabilities have already unlocked a new capability Google Cloud call “generative interfaces.” Google Cloud first two experiments, visual layout and dynamic view, represent this new type of interface, one generated by the model the moment you prompt it.

Visual layout generates an immersive, magazine-style view complete with photos and modules. These elements don't just look good; they also invite your input to further tailor the results. Ask Gemini to “plan a 3-day trip to Rome next summer,” and you'll get a visual itinerary you can actually explore.

Dynamic view takes a different approach, using Gemini 3’s agentic coding capabilities. Gemini designs and codes a custom user interface in real-time, perfectly suited to your prompt. For example, ask Gemini to “explain the Van Gogh Gallery with life context for each piece,” and you'll receive a stunning, interactive response that lets you tap, scroll and learn in ways static text cannot.

Google Cloud is rolling out visual layout and dynamic view today, and Google Cloud look forward to your feedback. To help us compare these experiments, you may initially see only one of them.

Gemini Agent: Multi-step tasks, handled

Gemini Agent is an experimental feature that handles multi-step tasks directly inside Gemini. It connects to your Google apps to manage your Calendar, add reminders, or just simply ask it to “organize my inbox" and it prioritizes to-dos and drafts replies for your approval. You can also give precise instructions, such as: "Research and help me book a mid-size SUV for my trip next week under $80/day using details from my email." Gemini will locate your flight information, compare rentals within budget and prepare the booking.

Built on insights from Project Mariner and powered by Gemini 3’s advanced reasoning, Gemini Agent breaks down complex requests using tools like Deep Research, Canvas, your Google Workspace connected apps like Gmail and Calendar, and live web browsing. When using it, you remain in control: Gemini is designed to seek confirmation before critical actions like making purchases or sending messages, and you can take over anytime. This marks Google Cloud next step toward a true generalist agent, and it will be available on the web for Google AI Ultra subscribers in the U.S. today.

All of today’s updates are driven by Google Cloud vision to create the most personal, proactive and powerful assistant on the planet.