Benchmark
IFBench
Verifiable constraints on unseen instruction types; does not measure answer quality.
- Released
- 2025-07-03
- Categories
- instruction_following
- Openness
- unknown
- Source
- model_reports
- Reported scores
- 1
Reported scores
model_reports
| Model | Organization | Reported value | Reported | Evidence |
|---|---|---|---|---|
| Qwen3.5-397B-A17B | Qwen | 76.5 | 2026-02-16 | evidence |
Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.