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.
Aggregate Rating 1798 ± 15, rank #141 of 2168 rated models, from 65 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| BALROG TextWorld (LLM) | 75.69 | Progress (%) | 100 |
| ComboShoppingBench - Budget Compliance | 87.3 | Pass rate (%) | 100 |
| ContractScrub | 0.66 | Macro F1 (0-1) | 100 |
| EYT-Bench (PersonaMem-v2) - Persona Alignment | 95.7 | Persona alignment (style, value, culture, memory, security) | 100 |
| LLMEval-Logic Formalization Free | 45.1 | Accuracy (%) | 100 |
| LLMEval-Logic Hard | 37.5 | Accuracy (%) | 100 |
| NoteVQA | 52.8 | Accuracy (%, thinking mode) | 100 |
| OCR-MetaReasoning | 89.3 | Meta-Reasoning Macro Score (%) | 100 |
| OCR-MetaReasoning - Data Interpretation | 96 | Answer Score (%) | 100 |
| OCR-MetaReasoning - Document Logic | 93 | Answer Score (%) | 100 |
| OCR-MetaReasoning - Field Dependency | 91.7 | Answer Score (%) | 100 |
| OCR-MetaReasoning - Meta-Abductive | 94.1 | Answer Score (%) | 100 |
Interactive version: theaggregate.ai/model?slug=gemini-3-1-pro-preview-thinking · How It Works · Data refreshed daily, snapshot 2026-10-20.