Google has released an upgraded version of Gemini 3 Deep Think, a specialized reasoning mode designed for complex scientific, research, and engineering challenges. This new iteration is available to Google AI Ultra subscribers and select researchers via the Gemini API. Early testers have used Deep Think to identify logical flaws in peer-reviewed papers, optimize crystal growth for semiconductor materials, and accelerate physical component design. The model has also achieved new benchmark records, including setting a new standard on Humanity's Last Exam and excelling in competitive programming and advanced scientific domains like physics and chemistry. AI
IMPACT Enhances capabilities in scientific and engineering problem-solving, potentially accelerating research and development.
RANK_REASON Google's direct announcement of an upgraded frontier model with new capabilities and benchmark results. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
Read on HN — AI startup stories →
- Anupam Pathak
- ARC-AGI-2
- ARC Prize Foundation
- Chemistry Olympiad
- CMT-Benchmark
- Codeforces
- Duke University
- Gemini 3 Deep Think
- Gemini API
- Google AI Ultra
- Humanity's Last Exam
- International Math Olympiad 2025
- International Physics Olympiad
- Liftware
- Lisa Carbone
- Rutgers University
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
