List your company

ResearchFormerly WizWand

HIDDENBENCH

Benchmarks

TaskDatasetSOTA resultTrendResults

Multi-agent split-evidence reasoning

HiddenBench clean run

33.8EM

6

Collective Reasoning

HIDDENBENCH 18 tasks

80Accuracy

6

Question Answering

HiddenBench

26.2Text EM

1