Gemini 2.5 Pro (High): benchmark results
Provider: Google. Released 2025-03-25. Access: API.
Aggregate Rating 1669 ± 17, rank #471 of 2257 rated models, from 78 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EpiCurveBench | 52.3 | EpiCurveSimilarity, ECS (%): edit distance with real penalty | 100 |
| EpiCurveBench - High Reasoning with Code Execution | 48 | EpiCurveSimilarity, ECS (%): edit distance with real penalty | 100 |
| EpiCurveBench - Manually Annotated Set | 56.9 | EpiCurveSimilarity, ECS (%): edit distance with real penalty | 100 |
| Misguided Attention - feeding_the_goat | 1 | Mean native expected-behavior score (0–1) | 94.6 |
| VTC-Bench (Interface Tools) - Perceptual Restoration | 40 | Pass rate (%) on perceptual restoration (50 questions under | 94.4 |
| Misguided Attention - lonely_rabbit_impossible | 1 | Mean native expected-behavior score (0–1) | 92 |
| VTC-Bench (Interface Tools) - Measurement | 44.76 | Pass rate (%) on measurement (105 questions on size, positio | 91.7 |
| Misguided Attention - bridge_torch_easy | 1 | Mean native expected-behavior score (0–1) | 91.1 |
| VTC-Bench (Code Tools) - Attention Focusing | 40 | Pass rate (%) on attention focusing (45 questions on rotated | 88.9 |
| VTC-Bench (Code Tools) - Robust OCR | 46 | Pass rate (%) on robust OCR (50 questions on degraded text), | 88.9 |
| VTC-Bench (Interface Tools) - Attention Focusing | 44.44 | Pass rate (%) on attention focusing (45 questions on rotated | 88.9 |
| VTC-Bench (Interface Tools) - Robust OCR | 60 | Pass rate (%) on robust OCR (50 questions on degraded text), | 88.9 |
Interactive version: theaggregate.ai/model?slug=gemini-2-5-pro-high · How It Works · Data refreshed daily, snapshot 2026-11-17.