Google is rolling out a major update to Gemini 3 Deep Think, a reasoning-focused mode now available to Google AI Ultra subscribers and select researchers via early API access. The model posts standout benchmarks — 48.4% on Humanity’s Last Exam, 84.6% on ARC-AGI-2, 3455 Codeforces Elo, and gold-level scores on the 2025 International Math Olympiad — and targets complex scientific and engineering tasks.
Key Points
- 1Achieves 48.4% Humanity's Last Exam, 84.6% ARC-AGI-2, 3455 Codeforces Elo, IMO gold
- 2Demonstrates deep multi-step reasoning and logical error detection in complex scientific contexts
- 3Enables engineers and researchers to verify models, prototype designs, and automate structured reasoning workflows
Scoring Rationale
Official Google release with industry-wide, actionable reasoning advances; limited by early restricted access and partly promotional claims.
Sources
Public references used for this report.
Practice with real Ad Tech data
90 SQL & Python problems · 15 industry datasets
Used by DS/ML engineers at top companies
Active Search Campaigns by BudgetEasyHigh CPC Clicks & Poor Landing PagesMediumCampaign ROAS by Attribution ModelHard
250 free problems · No credit card
See all Ad Tech problems
