Gemini 3.1 Pro (Preview): benchmark results
Google preview Gemini 3.1 Pro model for advanced reasoning and general tasks. Provider: Google. Released 2026-02-19. Access: API.
Aggregate Rating 1810 ± 2, rank #58 of 1656 rated models, from 4642 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| 3D-DefectBench - Geometry (Expert Labels) | 0.3 | Macro MCC (Matthews correlation from -1 to 1, macro-averaged | 100 |
| 3D-DefectBench - Texture (Silver Labels) | 0.16 | Macro MCC (Matthews correlation from -1 to 1, macro-averaged | 100 |
| AA Global-MMLU-Lite | 93.17 | Accuracy (%) | 100 |
| AA Global-MMLU-Lite - Arabic | 93 | Accuracy (%) | 100 |
| AA Global-MMLU-Lite - Bengali | 92.17 | Accuracy (%) | 100 |
| AA Global-MMLU-Lite - Burmese | 91.17 | Accuracy (%) | 100 |
| AA Global-MMLU-Lite - French | 93.67 | Accuracy (%) | 100 |
| AA Global-MMLU-Lite - German | 94.75 | Accuracy (%) | 100 |
| AA Global-MMLU-Lite - Hindi | 93.67 | Accuracy (%) | 100 |
| AA Global-MMLU-Lite - Indonesian | 93.67 | Accuracy (%) | 100 |
| AA Global-MMLU-Lite - Italian | 94.58 | Accuracy (%) | 100 |
| AA Global-MMLU-Lite - Japanese | 93.67 | Accuracy (%) | 100 |
Interactive version: theaggregate.ai/model?slug=gemini-3-1-pro-preview · How It Works · Data refreshed daily, snapshot 2026-10-20.