Google has unveiled Gemini 3, the latest and most advanced iteration of its AI model, now available to users worldwide across the Gemini app, Google Search AI Mode, AI Studio, Vertex AI, and Google Antigravity. Gemini 3 Pro launches in preview for consumers, developers, and enterprises, while Gemini 3 Deep Think mode is being tested by safety experts prior to release for Google AI Ultra subscribers. This rollout marks the first time a new Gemini model is launched in Search from day one, offering immediate access to advanced multimodal reasoning and agentic capabilities.
Gemini 3 introduces a host of technical improvements. It surpasses previous models and leading competitors on benchmarks such as:
- LMArena
- Humanity’s Last Exam
- GPQA Diamond
- MathArena Apex
- Vending-Bench 2
These benchmarks demonstrate PhD-level reasoning, long-horizon planning, and superior multimodal understanding across text, images, video, audio, and code. The model features a 1 million-token context window, leading multilingual performance, and reduced susceptibility to prompt injections and misuse. Gemini 3 Deep Think mode further raises the bar for complex reasoning and problem solving, exceeding Gemini 3 Pro’s scores on several evaluation metrics.
Google, through its DeepMind division, has rapidly iterated on Gemini since its initial launch two years ago. The company’s integrated approach combines research, infrastructure, and product delivery at scale, enabling a fast rollout to billions of users. With 2 billion monthly AI Overview users and over 650 million monthly Gemini app users, Google positions Gemini 3 as a foundational technology for learning, creation, and planning in both personal and enterprise contexts. Early testers and industry experts have highlighted Gemini 3’s step forward in reliability and the depth of reasoning capabilities.
