Gemini 3. Picture: Google AI
Google has launched the latest version of its artificial intelligence model, Gemini 3, emphasising that the new capabilities will be available immediately in several profit-generating products like its search engine.
Gemini 3, comes about eight months after Google introduced Gemini 2.5 and 11 months after Gemini 2.0 is likely to keep Google at the forefront of the AI race.
OpenAI, which kicked off the generative AI boom in late 2022 with the public launch of ChatGPT, introduced GPT-5 in August.
Intelligence
Google CEO Sundar Pichai described it as “our most intelligent model,” in a company blog post.
“The new AI model will allow users to get better answers to more complex questions, “so you get what you need with less prompting.”
The search giant said Gemini 3 is designed to bring state-of-the-art reasoning and powerful new agentic capabilities and will require users to do “less prompting” to get the desired results.
“Gemini 3 is set to transform the Google products people use every day, marking a significant leap forward in making AI truly helpful for everyone in South Africa and across the continent.”
Gemini 3 has arrived! 🚀
— Google (@Google) November 18, 2025
Here are three things to know about our most intelligent mode pic.twitter.com/6aM0ktQENI
South Africa
Kabelo Makwane, Country Director for Google South Africa, commented on the advantages of Gemini 3 for South Africa.
“For South Africa, the launch of Gemini 3 is more than a technological breakthrough – it is about democratising access to the most advanced intelligence and ensuring it translates directly into economic opportunity and job creation.
“This powerful new era of agentic AI is here, and it needs to work for us, helping our businesses thrive and empowering our people to acquire future-proof skills,” Makwane said.
NOW READ: AI momentum can solve problems including health and climate change, Dell says
Search trends
Makwane said the focus on localised impact is strongly supported by recent Google Search trends, which reveal that South African interest in career-focused AI learning has surged.
“With searches for “AI jobs” and “AI course” both having nearly doubled (+80%) in the past year, this signals a strong local appetite for tools that drive tangible professional and societal progress.|
Gemini 3 Pro, the first model in the new family, demonstrates PhD-level reasoning and sets new performance records on major AI benchmarks, from complex problem-solving (91.9% on GPQA Diamond) to understanding video, images, and even handwritten text.
Search responses
Beyond benchmarks, Gemini 3 Pro also brings a new quality to every interaction. Its responses are designed to be smart and direct, trading cliché for genuine insight to act as a true thought partner.
It excels at “vibe coding,” allowing developers to bring an idea to life with natural language, and its powerful agentic capabilities mean it can better understand intent, plan multi-step tasks, and get things done with less prompting from the user.
This enables smarter, more interactive experiences in Google Search, transforming the product from an information tool into an interactive learning companion.
Gemini
For the first time, a new Gemini model is available in Search from day one. In AI Mode, Gemini 3 creates a visual layout for a query, generating interactive tools and simulations on the fly.
Instead of just reading about physics, a user can now interact with a simulation of the three-body problem. Someone researching a mortgage can receive a custom-built loan calculator to compare options in real-time.
App
The Gemini app also features what Google called a powerful new agent which becomes a personal assistant that can take action on your behalf.
The Gemini app now has 650 million monthly active users and AI Overviews has 2 billion monthly users, the company said
For Google AI Ultra subscribers, the new experimental Gemini Agent can orchestrate and complete complex, multi-step tasks across connected Google apps.
A user can ask it to “organise my inbox,” and it will prioritise to-dos and draft replies. A request like, “Help me book a mid-size SUV for my trip,” allows the agent to find flight details from an email, compare rental options, and prepare the booking—all while keeping the user in control.
Antigravity
Google is also launching Google Antigravity, a new agentic development platform. This platform transforms AI from a simple coding assistant into an active partner that can autonomously plan and execute complex software tasks, from building features to fixing bugs, giving developers a powerful new way to bring their own ideas to life.
Google also announced a new agent platform called “Google Antigravity,” which lets developers code “at a higher, task-oriented level.”
“Google Antigravity is built on core tenets of trust, autonomy, and feedback. It is designed as a “mission control” for orchestrating multiple agents, providing developers with task-level context and verifiable artifacts to ensure they can trust the work being done.”
Availability
Google said this new agent-first approach represents a fundamental shift in how software will be built, allowing developers to operate at a higher, more strategic level.
Gemini 3 Pro is available to everyone starting today in the Gemini app and can be accessed by Google AI Pro and Ultra subscribers in AI Mode in Search.
NOW READ: Africa’s Gen Z’s reshaping the continent’s retail future
