Google Unveils Gemini 3, Its Most Intelligent Multimodal AI Model

Google announced the launch of Gemini 3, branding it as the company’s most intelligent and factually accurate AI system to date. The flagship Gemini 3 Pro model, available in the Gemini app and to select Search subscribers, is natively multimodal, handling text, images and audio together. Google highlighted new capabilities such as translating recipe photos, generating interactive flashcards, and creating visual, magazine‑style layouts. The rollout includes generative interfaces, an upgraded query‑fan‑out technique, reduced sycophancy, and an experimental Gemini Agent that can manage emails and book travel. The model is accessible to all users in the Gemini app, with additional features for Google AI Pro and Ultra subscribers.

Google Introduces Gemini 3 as Its Most Advanced AI Offering

Google has begun rolling out Gemini 3, a new series of AI models that the company describes as its “most intelligent” and “factually accurate” to date. The flagship version, Gemini 3 Pro, is being made available to everyone through the Gemini app on launch day and to subscribers inside Search. Google positions Gemini 3 as a leap forward that brings information "universally accessible and useful" for users across its ecosystem.

Native Multimodal Capabilities

Gemini 3 Pro is "natively multimodal," meaning it can process text, images and audio simultaneously rather than handling each modality separately. Google demonstrated practical uses such as translating photos of recipes into a full cookbook and generating interactive flashcards from a series of video lectures. These examples illustrate how the model can combine visual and textual data to produce richer, more actionable outputs.

Generative Interfaces and Visual Output

The new model powers "generative interfaces" that let users create visual, magazine‑style formats with pictures they can browse, as well as dynamic layouts tailored to specific prompts. Within the Gemini app, a built‑in workspace called Canvas enables users to build more "full‑feature" programs that leverage these visual capabilities. In Search’s AI Mode, Gemini 3 Pro can present results as images, tables, grids and simulations, enhancing the traditional text‑only experience.

Improved Search Techniques and Reduced Sycophancy

Google also upgraded its "query fan‑out" technique, allowing Gemini 3 Pro to break down complex questions into sub‑queries and better understand user intent. The company claims the model is less prone to empty flattery and exhibits "reduced sycophancy," delivering concise, direct insights rather than merely echoing what users want to hear.

Enhanced Reasoning and Agentic Features

Gemini 3 Pro brings stronger reasoning and longer‑horizon planning abilities, supporting more complex tasks. An experimental Gemini Agent feature lets the model act on behalf of users inside the Gemini app, handling actions such as reviewing and organizing emails or researching and booking travel. A "Deep Think" mode further boosts reasoning performance for safety testers.

Availability and Subscription Tiers

The model is now available inside the Gemini app for all users. Google AI Pro and Ultra subscribers in the United States can also try out Gemini Agent and access Gemini 3 Pro through AI Mode by selecting the "Thinking" option from the model dropdown. This tiered rollout aims to give a broad audience early access while offering advanced capabilities to paying subscribers.

Strategic Positioning

By launching Gemini 3, Google seeks to position itself ahead of competing AI providers, emphasizing factual accuracy, multimodal understanding and practical, user‑focused tools. The company frames the release as a step toward making information more universally useful across its suite of products.