SILO-BENCH

SILO-BENCH is a role-agnostic benchmark containing 30 algorithmic tasks across three communication complexity levels, used to evaluate the performance of distributed coordination in multi-agent large language model systems.

Papers using SILO-BENCH (2)

QueenBee Planner: Skill-Evolving Communication Topologies for Token-Efficient LLM Multi-Agent Systems2026Silo-Bench: A Scalable Environment for Evaluating Distributed Coordination in Multi-Agent LLM Systems2026