Hacker News · AI·· 1 d agoAI score66
Bespoke Labs Releases MAC Bench: 12 Tasks to Measure Multi-Agent Coordination
Multi Agent Coordination Benchmark
AI summary
Bespoke Labs has released the Multi Agent Coordination Bench (MAC Bench), which uses 12 native multi-agent tasks to separately measure the scores of a single agent, an eight-agent team, and eight agents without coordination, thereby separating the gains from coordination from the gains from adding more agents.
Source:Hacker News · AI · benchmarks.bespokelabs.ai