Google releases Gemini 3 Pro

Google DeepMind/Gemini

Google released Gemini 3 Pro, its most intelligent model, topping the LMArena leaderboard with a score of 1501. It also introduced Gemini 3 Deep Think, a slower reasoning mode given first to safety testers before becoming available to Google AI Ultra subscribers.

• Achieves PhD-level reasoning with 37.5% on Humanity's Last Exam without tools; Deep Think variant scores 41% and 45.1% on ARC-AGI-2.
• Excels on multimodal tasks: 81% on MMMU-Pro and 87.6% on Video-MMMU for visual reasoning.
• Available across Gemini app, Search AI Mode, Google AI Studio, Vertex AI, and the new Antigravity development platform.