Google began rolling out its next generation of artificial intelligence models, introducing Gemini 3 as the company’s “most intelligent” and “factually accurate” system to date. The launch marks a major step in Google’s push to regain momentum in the AI race following OpenAI’s troubled debut of GPT-5.
For the first time, Google is making its flagship Gemini 3 Pro model available to all users on day one through the Gemini app, while also integrating it into Search for subscribers.
Tulsee Doshi, senior director and head of product at Google DeepMind, said the new system advances the company’s mission of making information “universally accessible and useful,” especially as search evolves beyond text-only responses. “One really big step in that direction is to step out of the paradigm of just text responses and give you a much richer, more complete view of what you can actually see,” she noted.
Google describes Gemini 3 Pro as “natively multimodal,” capable of processing text, images and audio simultaneously. It can translate photographed recipes and convert them into a formatted cookbook, or generate interactive flashcards from video lectures.
The upgrades will be reflected across Google’s products. In the Gemini app, users can build more advanced projects using Canvas, the built-in workspace. Within Gemini Labs, the company is testing “generative interfaces,” which let the model present responses in magazine-style layouts or other interactive formats based on user prompts.
In AI Mode, the AI-powered version of Google Search, Gemini 3 Pro will surface more visual responses — including images, tables, grids and simulations. An improved “query fan-out” system is designed to better understand user intent and retrieve content previous models may have missed.
Google also drew a pointed contrast with OpenAI’s ChatGPT, saying Gemini 3 Pro delivers responses that are “smart, concise and direct,” with reduced “sycophancy” and fewer unnecessary pleasantries. Doshi said users will notice the model prioritising clarity and insight over flattery.
Gemini 3 Pro also features stronger reasoning and what Google calls “agentic” capabilities, enabling it to manage more complex tasks and plan over longer timeframes. An experimental Gemini Agent tool can review and organise emails or research and book travel directly within the app.
According to Google, Gemini 3 Pro now ranks at the top of LMArena, a widely used AI benchmarking platform. A new “Deep Think” mode further boosts its reasoning abilities, though it remains limited to safety testers for now.
