Global Launch of Google’s New Gemini: Google unveiled what it defines as a significant leap in the comprehension, analysis, and response capabilities of its AI systems – the Gemini 3 model. The model promises deep, accurate answers that are less dependent on precise prompt wording, and it is already integrated into the Gemini app, the search engine, and the company’s enterprise products.
Additionally, Google is introducing a new AI agents platform for developers called Antigravity, designed to create fully functional applications using short instructions. The announcement positions Google at the center of the race against major competitors in the field.
The competition between AI giants takes a new turn with Google’s announcement of the most advanced model it has developed so far. According to CEO Sundar Pichai, Gemini 3 is built to understand depth and nuance, with a stronger ability to grasp the contextual intent behind user requests.
Pichai wrote that the model allows users to get what they need with fewer instructions, shortening the path from question to high-quality answer. Google emphasizes that this is a significant upgrade over previous generations, launched only in recent months, reflecting the accelerated pace of development in the industry.
The new model will gradually be integrated into a wide range of services. AI-powered search applications will feature generative display interfaces capable not only of presenting answers but also of building full layouts including images, tables, frameworks, and interactive demonstrations.
Google showcases examples of requests generating designed articles, physical simulations, or interactive calculators built instantly. The company notes that the change is not just visual but based on a deeper understanding of the user’s original intent.
Alongside the consumer upgrade, Google is addressing developers directly: Antigravity, the new platform, enables coding based on broad tasks rather than individual lines of code. The platform allows developers to describe what they want to achieve conceptually, and the system builds the software structure for them.
Google describes this as progress in AI agent-based coding, currently referred to in the tech world as “ambient coding.” The company notes that Gemini 3 is the most precise model Google has developed for such tasks.
Google reveals that several search and AI Overview services already have billions of monthly users, presenting the new model to a massive audience. The company emphasizes that the model is designed to provide real insights rather than empty compliments.
DeepMind CEO Demis Hassabis stated that the system is designed to tell users what they need to hear, not what they want to hear, in response to critiques of earlier models seen as overly deferential.
The launch of Gemini 3 comes against the backdrop of two updates OpenAI introduced to GPT-5 recentlyk. Both companies are advancing rapidly, each aiming to provide users with deeper, more accurate, and reliable services. Google asserts that its recent developments place it at the forefront of computational capability, particularly in comprehension, analysis, and visual production.
For organizations, Google notes that the new model will be able to build employee training programs, analyze video documentation from production lines, detect faults, monitor employee movement and inventory, and fully automate business processes. Google’s vision is clear: To make the model a central work tool for companies of all sizes. The company adds that services will be gradually made available to all enterprise subscribers via Vertex AI.
With the broad launch, Google makes clear that the AI race has entered a new phase. The announcement coincides with increased investments by all major tech giants in computing infrastructure. Google, Meta, Microsoft, and Amazon are projected to invest more than $380 billion this year. All this indicates that models like Gemini 3 are not the final word but another stage in a long and demanding race.