List your company

ResearchFormerly WizWand

ProtocolBench

Benchmarks

TaskDatasetSOTA resultTrendResults

LLM Multi-Agent resilience-focused communication

ProtocolBench Fail-Storm scenario

14.91Pre-failure Discussion Rate

2

LLM Multi-Agent high-throughput serving

ProtocolBench Streaming Queue scenario

2,375Duration (s)

2

LLM Multi-Agent turn-based collaboration

ProtocolBench GAIA scenario

2.51Average Quality (1-5)

2

LLM Multi-Agent secure communication

ProtocolBench Safety scenario

—Primary metric

0