List your company

ResearchFormerly WizWand

MultiChallenge

Benchmarks

TaskDatasetSOTA resultTrendResults

Instruction Following

MultiChallenge

69.3Score

26

Instruction Following

MultiChallenge (Out-of-Domain)

38.5Overall Score

23

Reverse Chain-of-Thought Generation

MultiChallenge

45Score

20

General-purpose Behavior

MultiChallenge

58.6Score

7

Turn-level correlation with human ratings

MultiChallenge

0.74Spearman Correlation

6

Session-level correlation with human ratings

MultiChallenge

0.74Spearman Correlation (ρ)

6

Long-context & Multi-turn Dialogue

Multichallenge

39.71Score

4

Multi-turn Dialogue Reasoning

MultiChallenge

32.97Accuracy

4

Medical Instruction Following

MultiChallenge

66.8Pass@1

4

Relational Understanding

MultiChallenge

20.35IM Score

2