IFEval
Chat & Writing · 2023-11-14
IFEval measures instruction following using verifiable prompt constraints that can be checked automatically. The official Google Research repository is the direct provenance source.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude 3.5 Sonnet (Oct 2024) | 90.2 |
| Haiku 3.5 | 85.9 |