Gemini 3 Pro (High) — benchmark results
Gemini 3 Pro evaluated at the high reasoning-effort setting. Provider: Google. Released 2025-11-18. Access: API.
Unified ELO 1822 ± 26, rank #123 of 1776 rated models, from 39 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Vals AI MATH 500 | 96.4 | Accuracy (%) | 99.5 |
| MCPMark | 53.94 | Pass@1 (%) | 97.4 |
| VitaBench | 31.5 | Cross-Scenario Avg@4 (%) | 95 |
| Vals AI MGSM | 93.93 | Accuracy (%) | 90.9 |
| Pencil Puzzle Bench - Light Up | 20 | Direct-ask Success Rate (%) | 89 |
| Pencil Puzzle Bench - Slitherlink | 13.3 | Direct-ask Success Rate (%) | 88 |
| MedCode | 52.2 | Score (self-reported) | 86.8 |
| ALE-Bench | 1176.75 | Performance (Self-Refine x1) (self-reported) | 84.6 |
| IOI | 38.83 | Score (self-reported) | 77.4 |
| ATLAS | 62.6 | ATLAScore (self-reported) | 77.3 |
| Pencil Puzzle Bench - Nurimisaki | 6.7 | Direct-ask Success Rate (%) | 71 |
| Pencil Puzzle Bench - Shikaku | 6.7 | Direct-ask Success Rate (%) | 71 |
Interactive version: theaggregate.ai/model?slug=gemini-3-pro-high · How the rankings work · Data refreshed daily, snapshot 2026-07-22.