opt-13B: benchmark results
Provider: Meta. Released 2022-05-03. Access: Open.
Unified ELO 1259 ± 1, rank #1381 of 1392 rated models, from 8 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenBookQA | 39.8 | Accuracy (%) | 26.2 |
| HellaSwag | 69.9 | Accuracy (%) | 25 |
| WinoGrande | 64.7 | Accuracy (%) | 21.3 |
| BoolQ | 65 | Accuracy (%) | 18.2 |
| PIQA | 75.7 | Accuracy (%) | 16.7 |
| ARC Challenge (AI2) | 35.8 | Accuracy (%) | 11.8 |
| ChineseSafe Benchmark | 50.18 | Accuracy (%) | 3.7 |
| MMLU | 25.1 | Accuracy (%) | 0 |
Interactive version: theaggregate.ai/model?slug=opt-13b · How It Works · Data refreshed daily, snapshot 2026-09-05.