Instruction Following Evaluation from the Open LLM Leaderboard — tests precise adherence to complex, multi-constraint instructions.
Standardized IFEval scoring across many open models enables reliable cross-model comparison; explicit constraint checks make failures easy to diagnose and reproduce.
Retired by Hugging Face on March 13, 2025; retained scores are historical. Focuses on explicit constraints and under-tests ambiguous intent resolution; high scores do not guarantee factual or domain-correct outputs.
Higher is better. Percentage accuracy.
Max score: 100
Retired by Hugging Face on March 13, 2025.
| # | Model | Score |
|---|---|---|
| 1 | 90.0 | |
| 2 | 90.0 | |
| 3 | 90.0 | |
| 4 | 87.4 | |
| 5 | 86.9 | |
| 6 | 86.7 | |
| 7 | 86.7 | |
| 8 | 86.5 | |
| 9 | 86.4 | |
| 10 | 84.1 | |
| 11 | 84.0 | |
| 12 | 83.5 | |
| 13 | 82.0 | |
| 14 | 81.3 | |
| 15 | 81.0 | |
| 16 | 80.9 | |
| 17 | 80.4 | |
| 18 | 80.2 | |
| 19 | 79.9 | |
| 20 | 79.9 | |
| 21 | 79.8 | |
| 22 | 78.6 | |
| 23 | 78.2 | |
| 24 | 78.0 | |
| 25 | 78.0 | |
| 26 | 77.8 | |
| 27 | 77.8 | |
| 28 | 77.7 | |
| 29 | 77.6 | |
| 30 | 77.5 | |
| 31 | 77.4 | |
| 32 | 77.1 | |
| 33 | 77.1 | |
| 34 | 77.0 | |
| 35 | 76.9 | |
| 36 | 76.9 | |
| 37 | 76.8 | |
| 38 | 76.6 | |
| 39 | 76.4 | |
| 40 | 76.3 | |
| 41 | 76.2 | |
| 42 | 76.0 | |
| 43 | 76.0 | |
| 44 | 75.9 | |
| 45 | 75.8 | |
| 46 | 75.7 | |
| 47 | 75.6 | |
| 48 | 75.2 | |
| 49 | 75.1 | |
| 50 | 75.1 |