Google launches Gemini 3

New model moves to the top of the LMArena Leaderboard.

Author

Innovation Profs
November 21, 2025

đź“« Subscribe here | 🗣️ Hire us to speak | 🎤 See upcoming events

Google launches a Gemini update

Move over ChatGPT, Claude and Copilot. Google has entered the chat.

Google this week launched Gemini 3, it's “most intelligent” and “factually accurate” AI model yet, with big jumps in reasoning, coding and multimodal understanding compared to Gemini 2.5 Pro.

This new model is available to everyone in the Gemini app and inside search for paying subscribers.

The model is natively multimodal, handling text, images, and audio together, and powers new features like richer “Dynamic View” search answers, a Canvas workspace for building more complete apps, and experimental “generative interfaces” that turn answers into magazine style layouts.

Gemini 3 Pro focuses on better coding and reasoning, plus what Google pitches as less flattery and more direct, useful responses. It also includes a new Gemini Agent that can handle multi-step tasks like organizing email or researching and booking travel. The model is already topping popular community benchmarks, and a more intensive Deep Think mode is in testing for even stronger reasoning.

Sign up for our free AI update & virtual Copilot training

HOLID-AI 3: Join us at noon Central on Dec. 12 for our third annual HOLID-AI generative AI update. We’ll catch you up on all the latest news in the world of generative AI - from AI browsers to LLM updates to video advancements and more. Sign up here. The event is free.

Virtual Microsoft Copilot Workshop: This three-hour virtual workshop on Dec. 16 will show you how to get the most out of Microsoft Copilot. You’ll learn how to use the LLM more efficiently, how to use and build Agents and how to integrate Copilot tools into apps such as Word, Excel, Outlook, Teams and more. Sign up here to get the early bird rate (and use the code BLACKFRIDAY to save $25 before Nov. 30).

Google’s image update

Google also this week announced the release of Nano Banana Pro (also known as Gemini Pro Image), an update to its image generation tool that took the AI world by storm in late August. New features let users:

  • Generate clearer text within images

  • Seamlessly transfer image styles

  • Combine 6 images at a time with high-quality results

  • Turn ideas into professional-quality visual.

Nano Banana Pro is available now in Gemini and Google AI Studio.

Here Nano Banana Pro turned me into an NBA player:

And to promote our upcoming Copilot workshop:

More new models

Gemini 3 wasn't the only new model released this week.

OpenAI released GPT 5.1, an update to GPT 5, that it described as smarter and more conversational. It comes with two models, which it describes as:

  • GPT‑5.1 Instant: our most-used model, now warmer, more intelligent, and better at following your instructions.

  • GPT‑5.1 Thinking: our advanced reasoning model, now easier to understand and faster on simple tasks, more persistent on complex ones.

Meanwhile, Grok released a new model Grok 4.1. It's available on grok.com, the X platform, and the Grok mobile apps. The company describes Grok 4.1 as "more perceptive to nuanced intent, compelling to speak with, and coherent in personality, while fully retaining the razor-sharp intelligence and reliability of its predecessors."

How they rate

Gemini 3 took over the top spot for writing text on the LMArena Leaderboard. It is followed closely by both Grok 4.1 thinking and Grok 4.1.

OpenAI's GPT 5.1 High comes in at fourth.