Gemini 3.7 Flash (Medium): benchmark results
Provider: Google. Released 2026-08-13. Access: API.
Unified ELO 1728 ± 1, rank #54 of 1761 rated models, from 25 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA-LCR | 81 | Accuracy (self-reported) | 98.3 |
| AA Omniscience - Law | 60.1 | Accuracy (%) | 97.3 |
| AA Long Context Reasoning | 83 | Accuracy (%) | 97 |
| AA MMMU-Pro | 84.74 | Accuracy (%) | 96.7 |
| AA SciCode | 59.84 | Accuracy (%) | 96.6 |
| Epoch AI - Scicode | 57.87 | Score | 95.6 |
| AA Omniscience - Humanities & Social Sciences | 53.7 | Accuracy (%) | 94.2 |
| AA-Omniscience Accuracy | 54 | Accuracy (%) | 94.2 |
| AA Omniscience | 23.7 | Score | 94 |
| AA Omniscience - Business | 45.3 | Accuracy (%) | 94 |
| AA GPQA Diamond | 92.12 | Accuracy (%) | 93.7 |
| AA Omniscience - Science, Engineering & Mathematics | 51.1 | Accuracy (%) | 93.5 |
Interactive version: theaggregate.ai/model?slug=gemini-3-7-flash-medium · How It Works · Data refreshed daily, snapshot 2026-09-05.