Google DeepMind announced Gemini 3 and Gemma 3, marking major progress in AI reasoning and multimodal capabilities. Gemini 3 Pro achieved 23.4% on MathArena Apex, setting a new benchmark for frontier math models, while excelling on Humanity's Last Exam and GPQA Diamond. This leap enables more complex problem-solving across science and coding.

The Gemma 3 family expands access with lightweight, open models capable of running on single GPUs or TPUs. Improvements include expanded multilingual support and a larger context window, making advanced AI usable for diverse developers and researchers globally. These models underpin innovations in Google's products.

AI integration now powers Pixel 10 features and transformed Search with agentic capabilities. New tools like Google Antigravity revolutionize software development by acting as collaborative coding partners. Creative tools gained Veo 3.1 for video and Nano Banana Pro for image editing, enhancing user creativity across Google's ecosystem.

Beyond products, Gemini 3 accelerates quantum computing research and aids genomics breakthroughs. Google emphasizes responsible development through initiatives addressing climate change and global challenges, collaborating widely to turn lab innovations into real-world solutions impacting daily life.