Receive daily AI-curated summaries of engineering articles from top tech companies worldwide.
Endigest AI Core Summary
Google releases a major upgrade to Gemini 3 Deep Think, its specialized reasoning mode designed for science, research, and engineering challenges.
•Sets a new standard on Humanity's Last Exam with 48.4% (without tools), surpassing previous frontier model benchmarks
•Achieves 84.6% on ARC-AGI-2, verified by the ARC Prize Foundation, and an Elo of 3455 on Codeforces
•Demonstrates gold-medal performance on the 2025 International Math Olympiad, Physics Olympiad, and Chemistry Olympiad written sections
•Real-world testers include Rutgers University mathematician who used it to detect a logical flaw missed by human peer review, and Duke's Wang Lab which used it to design thin film growth recipes exceeding 100 μm
•Now available to Google AI Ultra subscribers and via the Gemini API through an early access program for researchers and enterprises
This summary was automatically generated by AI based on the original article and may not be fully accurate.
Deepmind
Google Cloud
Databricks
Stripe