Google announced, this Tuesday (18), the launch of the Gemini 3, its most advanced artificial intelligence model to date. Disponível out of the box in the Gemini app and Modo AI from Pesquisa Google, the new version promises more concise responses, greater programmability, and interactive experiences through generative interfaces. The objective is to offer a versatile tool for users and developers, with direct integration into the company’s products. The simultaneous release of the chatbot and the search engine marks a first in the company’s strategy.
The technology was developed by Google DeepMind and stands out for its advances in multimodality, allowing it to process text, images, videos and codes with greater precision. Segundo the company, the Gemini 3 achieved superior results in reasoning and programming benchmarks. The new feature is now accessible globally for users over 18 years old in countries where the application is supported.
- Key innovations: Generative Interfaces, multimodal understanding, optimized search.
- Availability: Aplicativo Gemini, Modo AI from Pesquisa Google and Vertex AI.
- Target audience: general Usuários, developers and companies via Google Cloud.
Advances in generative interfaces
Gemini 3 introduces generative interfaces, capable of creating visual platforms in real time. The “Visual Layout” feature organizes responses with images, videos and tables, simulating a specialized website experience.
The “Dynamic View” functionality allows you to generate complete interfaces, such as web applications, from simple prompts. Um example presented was the creation of an interactive gallery about Vincent van Gogh, with historical context and images of works, generated in seconds.
Improved multimodal understanding
The new AI processes different media formats with greater efficiency. Testes benchmarks show superior performance on tasks that combine text, image and video.
For example, the Gemini 3 can extract detailed annotations from images or adapt responses to specific formats, such as visual presentations. Essa capability is useful for students and professionals dealing with complex content.
Integration with development tools has also been optimized. The model supports the creation of more accurate codes, benefiting programmers and companies.
New look of the application
The Gemini application has a renewed interface, with more intuitive navigation. The “My Stuff” section makes it easy to access created content, such as images and reports.
The redesign aims to improve the user experience, especially on mobile devices. The update is now available for download in app stores.
Smarter programming
The Gemini 3 reinforces the concept of “vibe coding”, with greater precision in programming tasks. The model is capable of generating complex codes, such as SVGs, outperforming competitors in benchmarks.
Companies using Vertex AI will have access to advanced tools for automation and development. Integration with Google Cloud expands use in corporate projects.
More efficient Google search
For the first time, a new AI model was integrated into Pesquisa Google on launch day. The Modo IA now offers more visual and interactive responses, with tables and graphs.
Subscribers to the Pro and Ultra plans have immediate access to the “Thinking” feature, which uses the Gemini 3 to process complex questions. The functionality will be expanded to all users soon.
Integration with Google Antigravity
Google Antigravity, a new development platform, accompanies the launch. Ela allows programmers to create applications with high-level commands, integrating AI into the process.
The tool combines text prompts, terminal, and real-time preview. It is aimed at developers who seek agility in creating software.