MAC Bench finds multi-agent teams underperform single agents
Bespoke Labs released MAC Bench, 12 long-horizon tasks measuring coordination among AI agents. Surprisingly, teams of 8 agents underperformed a single agent, exposing weak multi-agent collaboration in frontier models.
2026-11-19 ~ 2026-11-19 · 3 related posts
- New MAC Bench finds frontier agents coordinate poorly: 8-agent teams win by exploration, not teamwork — AlexGDimakis · 2026-11-19
- Multi Agent Coordination Bench Shows Agents Benefit From Randomness, Not Teamwork — soldni · 2026-11-19
1 near-duplicate retellings: AlexGDimakis