opt-iml-max-1.3B: benchmark results
Meta OPT-IML-Max is the instruction-tuned OPT release trained on approximately 2,000 tasks from eight benchmark collections. The Max training mix remains distinct from OPT-IML, which reserves tasks for evaluation, and from the pretrained OPT base model. Provider: Meta. Access: Open.
Aggregate Rating 1156 ± 17, rank #9863 of 10916 rated models, from 68 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MMLU-by-task - College Physics | 34.31 | Accuracy (%) | 86.4 |
| MMLU-by-task - High School Statistics | 46.3 | Accuracy (%) | 78.9 |
| MMLU-by-task - High School Mathematics | 26.3 | Accuracy (%) | 40 |
| MMLU-by-task - Human Sexuality | 34.35 | Accuracy (%) | 35.5 |
| MMLU-by-task - Management | 37.86 | Accuracy (%) | 35.3 |
| MMLU-by-task - High School Government And Politics | 38.34 | Accuracy (%) | 35.2 |
| MMLU-by-task - High School Biology | 32.9 | Accuracy (%) | 35.1 |
| MMLU-by-task - US Foreign Policy | 41 | Accuracy (%) | 35.1 |
| MMLU-by-task - High School World History | 35.86 | Accuracy (%) | 35 |
| MMLU-by-task - Nutrition | 32.68 | Accuracy (%) | 33.9 |
| MMLU-by-task - Clinical Knowledge | 32.45 | Accuracy (%) | 33.7 |
| MMLU-by-task - High School European History | 29.7 | Accuracy (%) | 32.8 |
Interactive version: theaggregate.ai/model?slug=opt-iml-max-1-3b · How It Works · Data refreshed daily, snapshot 2026-11-08.