Gemini 3.1 Pro (Preview) (Thinking) — benchmark results
Gemini 3.1 Pro (Preview) evaluated with thinking enabled. Provider: Google. Released 2026-02-19. Access: API.
Unified ELO 1907 ± 30, rank #61 of 1776 rated models, from 14 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| BALROG TextWorld (LLM) | 75.7 | Progress (%) | 100 |
| LLMEval-Logic Formalization Free | 45.1 | Accuracy (%) | 100 |
| LLMEval-Logic Hard | 37.5 | Accuracy (%) | 100 |
| MathVision | 89.8 | Overall Accuracy (%) | 97.2 |
| BALROG BabyAI (LLM) | 98 | Progress (%) | 97 |
| BALROG BabaIsAI (LLM) | 83.3 | Progress (%) | 93.9 |
| BALROG Crafter (LLM) | 55 | Progress (%) | 92.4 |
| LLMEval-Logic Hard Sub-Q | 76 | Accuracy (%) | 92.3 |
| BALROG NetHack (LLM) | 2.6 | Progress (%) | 90.9 |
| BALROG MiniHack (LLM) | 27.5 | Progress (%) | 86.4 |
| LLMEval-Logic Base | 74 | Accuracy (%) | 84.6 |
| SEAL - SWE-Bench Pro (Public Dataset) | 46.1 | Score | 83.3 |
Interactive version: theaggregate.ai/model?slug=gemini-3-1-pro-preview-thinking · How the rankings work · Data refreshed daily, snapshot 2026-07-22.