Google has unveiled Gemini 3, its most advanced artificial intelligence model yet, introducing a new era of deep reasoning, intelligent agent capabilities, and next-generation assistance across its products.
The rollout, which begins today for select users, marks one of the biggest leaps in Google’s AI evolution in recent years.
Okay News reports that the company is positioning the new model as the backbone of a more intuitive, more proactive, and more personalised AI experience for billions of people worldwide.
Sundar Pichai, CEO of Google and Alphabet, described the launch as “a new chapter” in the company’s effort to build AI that is truly helpful to everyone. He noted that interest in AI-powered learning tools is accelerating rapidly across several countries, including Nigeria, where Google Search queries on “AI and studying” have surged by over 200 percent in the past year, alongside a 160 percent increase in searches for “how to learn AI”.
At the heart of the announcement is Gemini 3 Pro, the first model in the new family and the most capable system Google has ever built. The company says it demonstrates PhD-level reasoning and has set new performance records across major global benchmarks.
Gemini 3 Pro achieved 91.9 percent on the GPQA Diamond reasoning test and shows improved understanding of video, images, and even handwritten documents. Google emphasised that these advances translate into clearer, more direct, and more insightful responses during everyday use, trading generic AI phrasing for what it described as “genuine thought partnership”.
A major shift is also coming to Google Search, where Gemini 3 is now embedded from day one. Instead of simply returning links or text, the AI can now build interactive tools and visual explanations on the fly.
Users searching for complex physics problems will see simulations they can manipulate, while someone researching mortgage plans may instantly receive a custom loan calculator tailored to their query. Google says this move transforms Search into an interactive learning companion rather than a static information source.
The Gemini app is also gaining a powerful upgrade with the introduction of an experimental Gemini Agent, available to Google AI Ultra subscribers. The agent is capable of completing multi-step tasks across connected Google apps. A request such as “organise my inbox” will prompt it to sort messages, highlight priorities, and draft replies. A task like “plan my trip and find an SUV to rent” allows the agent to pull flight details from an email, compare rental options, and prepare a booking — while leaving final approval with the user. Google says the agent is designed to support everyday tasks without compromising user control.
To help developers build experiences powered by these new capabilities, Google also introduced Google Antigravity, an agentic development platform designed to plan and execute complex software tasks autonomously. The company described Antigravity as a “mission control” system where multiple AI agents can be coordinated, allowing developers to work at a higher, more strategic level while the platform handles code generation, feature planning, debugging, and system optimisation.
Google stressed that Gemini 3 has undergone its most comprehensive safety reviews to date. The model is available beginning today in the Gemini app and will also power AI Mode in Search for Google AI Pro and Ultra subscribers.
With the arrival of Gemini 3, Google says it is entering a new phase of AI development — one built on deeper intelligence, stronger agent capabilities, and personalised assistance designed to help users accomplish more with less effort.
