Google's new artificial intelligence model: Gemini 3
Sait Çıkrıkçı
19.11.2025
169
Google executives introduced Gemini 3, the new version of the Gemini AI family, noting that it represents a leap forward in the model's reasoning and coding capabilities compared to previous versions.
Gemini 3 becomes instantly available across all of Google's core services, including its search product, while also gaining the ability to answer questions with interactive graphics.
The company stated that, like the previous version of the model, it can process text, images, and various media types, and that its ability to solve complex science and math problems has also been enhanced.
Google CEO Sundar Pichai said in a statement on Tuesday that the model's ability to reason and respond based on these inputs has significantly improved. "In just two years, artificial intelligence has evolved from reading text and images to reading a room," Pichai said.
Google executives stated that the model can convert different types of information between formats and generate images or applications with a single command. For example, when asked for a travel plan, Gemini 3 can create images with clickable interactive components.
Google Gemini team leader Josh Woodward said, "It's not just about how Gemini 3 understands input. It's also important how it can generate outputs in entirely new ways."
The new model will be used to answer the most complex questions in Google Search or AI Mode, while simpler questions will be directed to other Gemini models.
New development platform Antigravity
Google also introduced Antigravity, a new development platform for creating AI-powered autonomous coding agents. Currently in preview, the system allows developers to assign tasks to autonomous agents that can write, test, and verify code in an editor, terminal, and browser.
Managers said that when a developer wants to bring an idea to life, such as creating a flight tracking application, the agent can complete the task by working simultaneously across tools.
"Gemini 3 Deep Think" was also introduced
Google also announced the Gemini 3 Deep Think feature alongside the new model. This mode offers an advanced reasoning system that selects the most suitable solution by testing multiple hypotheses in parallel.
It is stated that it will deliver high performance in complex tasks requiring coding, scientific research, and multi-step planning. Deep Think will initially be available in Google's highest-priced AI subscription, the Google AI Ultra plan, priced at $249.99 per month.
Source : bloomberght.com
The company stated that, like the previous version of the model, it can process text, images, and various media types, and that its ability to solve complex science and math problems has also been enhanced.
Google CEO Sundar Pichai said in a statement on Tuesday that the model's ability to reason and respond based on these inputs has significantly improved. "In just two years, artificial intelligence has evolved from reading text and images to reading a room," Pichai said.
Google executives stated that the model can convert different types of information between formats and generate images or applications with a single command. For example, when asked for a travel plan, Gemini 3 can create images with clickable interactive components.
Google Gemini team leader Josh Woodward said, "It's not just about how Gemini 3 understands input. It's also important how it can generate outputs in entirely new ways."
The new model will be used to answer the most complex questions in Google Search or AI Mode, while simpler questions will be directed to other Gemini models.
New development platform Antigravity
Google also introduced Antigravity, a new development platform for creating AI-powered autonomous coding agents. Currently in preview, the system allows developers to assign tasks to autonomous agents that can write, test, and verify code in an editor, terminal, and browser.
Managers said that when a developer wants to bring an idea to life, such as creating a flight tracking application, the agent can complete the task by working simultaneously across tools.
"Gemini 3 Deep Think" was also introduced
Google also announced the Gemini 3 Deep Think feature alongside the new model. This mode offers an advanced reasoning system that selects the most suitable solution by testing multiple hypotheses in parallel.
It is stated that it will deliver high performance in complex tasks requiring coding, scientific research, and multi-step planning. Deep Think will initially be available in Google's highest-priced AI subscription, the Google AI Ultra plan, priced at $249.99 per month.
Source : bloomberght.com
Comments
Add Comment
No comments yet. Be the first to comment!
