Google unveils upgraded Gemini 3 AI model as competition intensifies

SAN FRANCISCO: Google on Tuesday rolled out the newest version of its Gemini artificial intelligence system, introducing enhanced capabilities across its dedicated app and flagship search engine as the company pushes to strengthen its position in the rapidly evolving AI race.

The upgraded model, Gemini 3, has been described by its creators as the company’s most advanced AI yet — capable of interpreting and analysing text, images, audio and video, and powering a new generation of digital agents that can build applications on command.

“This is our most intelligent model,” Google’s head of AI Koray Kavukcuoglu told reporters during a briefing. “We believe it will help anyone bring their ideas to life.”

Google said the new model will be available to more than 650 million monthly users of the Gemini app, and to over two billion monthly users on Google Search through the Gemini-powered Overviews feature.

Kavukcuoglu said the company has accelerated its AI rollout strategy over the past year.

“Gemini has set a new pace—both in releasing models and getting them to users faster than ever. We’ve been shipping rapidly and learning from that experience.”

Google’s renewed momentum comes after a rocky start when rivals, particularly OpenAI, surged ahead with ChatGPT. Early setbacks—including widely mocked generative AI errors — left Google scrambling to regain its footing. But since debuting the first version of Gemini in late 2023, the tech giant has staged a significant comeback in consumer-facing AI.

The pace of development remains relentless. Gemini 3 is being introduced just nine months after the previous iteration, signalling Google’s determination to stay competitive as OpenAI continues to refine its own models and defend its lead in the sector.

Big tech firms are meanwhile pouring tens of billions of dollars into AI infrastructure, fuelling concerns about the sustainability of investments that have driven a historic stock market surge.

Google CEO Sundar Pichai, in an interview with the BBC, acknowledged “irrationality” in the current wave of AI spending and warned of the “immense” energy demands of advanced systems. He also urged users not to “blindly trust” AI responses, stressing the importance of complementary tools like traditional search.

Google executives say Gemini 3 marks a major leap in reasoning and problem-solving.

“This is a huge step forward,” said vice president of product Robby Stein. “You’ll now be able to get answers to some of your hardest questions directly in search.”

According to Tulsee Doshi, senior director of product management, Gemini 3 handles natural language with more depth, precision and nuance than earlier versions.

Kavukcuoglu added that responses are crafted to be “clever, concise and direct,” while avoiding clichés and unnecessary flattery.

Google also announced Antigravity, a new software development platform powered by an autonomous AI agent capable of writing code based on plain-language instructions—part of the company’s push to integrate AI deeper into development workflows.

The updated Gemini model has also been reinforced with stronger security protections, designed to make it more resistant to cyberattacks, prompt manipulation and other vulnerabilities.