Alphabet, Google's parent company, launched Gemini 3, its new generation of artificial intelligence model, on Tuesday, aiming to compete head-on with OpenAI in the field of generative AI. Dubbed the most advanced model developed by Alphabet, Gemini 3 stands out for its ability to provide more accurate and nuanced responses, while requiring fewer instructions. CEO Sundar Pichai hailed an AI that is "useful rather than flattering," capable of better understanding the user's intentions. The model will be gradually integrated into Google's AI search products, the Gemini app, which has 650 million monthly users, and professional services via Vertex AI and the Gemini API.
Google also unveiled Antigravity, a new task-oriented development platform designed for natural language assisted coding. According to Google Labs, Gemini 3 offers a "vibe coding" experience, allowing users to generate interfaces, visualizations, and interactive explanations, similar to a "digital magazine." The tool targets both developers and businesses, offering a variety of uses: creating interactive simulators, analyzing industrial videos, or generating HR content such as onboarding modules.
The launch comes at a time of rapid acceleration in investment in artificial intelligence. The digital giants Alphabet, Microsoft, Amazon, and Meta are expected to collectively spend over $380bn this year on their AI infrastructures. Gemini 3 is thus positioned as a direct response to recent developments in GPT-5 at OpenAI, while signaling Google's desire to make its models more assertive and less consensual. With Gemini 3, Google intends to reestablish its technological leadership and spread AI "across its entire infrastructure," Sundar Pichai says.
