SEOUL/MOUNTAIN VIEW – Google escalated the artificial intelligence arms race on November 18th (local time) by announcing the launch of its next-generation AI model, Gemini 3. This release comes just eight months after Gemini 2.5 and a mere three months following OpenAI's debut of GPT-5, signaling a renewed, fierce battle for market dominance in the rapidly evolving AI landscape.

Sundar Pichai, CEO of Alphabet (Google's parent company), announced the news via a blog post. "Gemini 3 is engineered to deliver the desired answer to complex questions with minimal prompting," Pichai stated, adding, "We are now delivering Gemini at Google's scale." Prompting, in this context, refers to the questions, instructions, or commands given to the AI.

Broad Integration and Enhanced Capabilities 

Google is set to integrate Gemini 3 across its core product lineup, including the Gemini app, AI Mode, AI Overview, and enterprise services. The rollout began immediately for some paid subscribers and is expected to expand to general users within the coming weeks.

The scale of Google's AI reach is substantial. The Gemini app currently boasts 650 million monthly active users, while AI Overview—Google's AI-powered summarization and organization feature in search results—is utilized by a staggering 2 billion users monthly. This rivals OpenAI's user base, which reported 700 million weekly ChatGPT users in August, underscoring the clear and intensifying competitive dynamic between the tech giants.

Pichai emphasized that Gemini 3 features significantly enhanced capabilities in grasping context and user intent. He explained that the model has been "upgraded to understand deeper and more nuanced requests."

Demis Hassabis, CEO of Google's AI research organization, DeepMind, highlighted a key design philosophy: "Gemini 3 is engineered to present information that users genuinely need, rather than offering generic praise or formulaic responses." This remark is widely seen as an acknowledgment of recent industry critiques regarding AI chatbots' tendency toward overly flattering or subservient interactions. DeepMind, founded in the UK in 2010 and acquired by Google in 2014, is the core research division responsible for Google's large-scale AI model and foundational research, having previously gained global renown for developing the Go-playing AI, AlphaGo.

Targeting the Developer Market with New Tools 

Alongside Gemini 3, Google introduced a new development platform called Google Antigravity. This tool is designed to automate code generation at the work unit level, directly targeting the burgeoning "prompt-based coding" (often called "vibe coding") market. Josh Woodward, Vice President of Google Labs, stressed that Gemini 3 possesses "the most outstanding code generation capability of any model Google has developed."

Furthermore, Gemini 3 moves beyond simple text-based responses by offering a generative interface that incorporates images, tables, and grids. For example, a request to "explain a Van Gogh gallery with background context on the artwork" could generate visual content in the style of a digital magazine. In AI Mode, Gemini 3 supports various visual applications, such as creating interactive loan calculators and configuring complex physics problem simulations.

For enterprise customers, Gemini 3 promises substantial operational improvements, including automatic generation of employee training materials, enhanced precision in video analysis, factory image analysis, and procurement automation. Developers can integrate Gemini 3 via the Gemini API, while businesses can leverage Google Cloud's Vertex AI platform.

The intensifying AI race is reflected in massive infrastructure investment across the industry. Google, Meta, Microsoft, and Amazon are projected to spend over $380 billion combined in capital expenditures this year. OpenAI, not to be outdone, announced consecutive performance updates for its GPT-5 model last week, ensuring the competition remains at a fever pitch.