Google Gemini 3 Deep Think - Humanity's Last Exam (search and code) — leaderboard
Metric: Score (%). Source: storage.googleapis.com. 4 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Deep Think | 53.4 |
| 2 | Claude Opus 4.6 (Thinking) | 53.1 |
| 3 | Gemini 3 Pro (Preview) | 45.8 |
| 4 | GPT-5.2 (xHigh) | 45.5 |
Interactive version: theaggregate.ai/benchmark?slug=google-gemini-3-deep-think-humanity-s-last-exam-search-and-code · How the rankings work · Data refreshed daily, snapshot 2026-07-22.