MMLU
Measuring Massive Multitask Language Understanding. A test to measure a text model's multitask accuracy.
State-of-the-Art Leaderboard
| Rank | Model | Metric | Score | Paper | Date |
|---|---|---|---|---|---|
| 🏆 1 | Claude Opus 5.5 | Accuracy | 89.2% | - | 2026-09-10 |
| 2 | GPT-6 Astra | Accuracy | 88.9% | GPT-4 Technical Report | 2026-09-01 |