Trending 
Google unveiled a new version of its Gemini artificial intelligence model on Tuesday, rolling it out across its dedicated Gemini app and integrating it more deeply into its widely used search engine as the company intensifies its push to lead the global AI race. The updated model, Gemini 3, was promoted by its developers as the most advanced system yet for analysing and generating responses from text, images, audio and video, while also powering a sophisticated new digital “agent” capable of building applications on demand.
Describing the release to journalists, Google’s head of AI, Koray Kavukcuoglu, called it the company’s “most intelligent model”, expressing confidence that it would enable users to bring their ideas to life more easily than ever. The upgraded AI will be accessible in the Gemini app, which now has over 650 million monthly users, and via Google Search, where Gemini powers the Overviews feature used by more than two billion people each month.
Kavukcuoglu noted that the rapid rollout of new Gemini versions marks a shift in Google’s development pace, following the company’s early missteps in the generative AI space when it was overshadowed by OpenAI’s ChatGPT. Since launching the first version of Gemini in late 2023, Google has repositioned itself as a major competitor in consumer AI, although its AI Overviews feature has occasionally faced mockery for producing unusual or inaccurate responses.
The debut of Gemini 3, arriving just nine months after its predecessor, reflects Google’s determination to keep up momentum as competition heats up with OpenAI and other rivals. Big technology firms have been pouring tens of billions of dollars into AI development, although growing investor scepticism has begun to temper Wall Street’s earlier enthusiasm.
Google chief executive Sundar Pichai acknowledged in an interview with the BBC that AI investment has shown signs of “irrationality”, fuelling a historic surge in tech valuations. He also warned of the significant energy demands associated with AI systems and advised users not to “blindly trust” AI-generated answers, arguing that traditional tools such as search engines still play an important role alongside chatbots and AI summaries.
Robby Stein, Google’s vice-president of product, said the addition of Gemini 3 marked a substantial leap forward, with the model delivering advanced reasoning capabilities directly within search. Tulsee Doshi, senior director of product management, added that the system now handles natural language queries with greater depth and subtlety. Kavukcuoglu stressed that the aim was to provide responses that were clear, thoughtful and free of clichés.
Google also announced Antigravity, a new software development platform powered by an AI agent capable of autonomously writing code from simple instructions. The Gemini team said security had also been fortified in the latest version, making it more resilient against cyber threats, prompt manipulation and other vulnerabilities.
