New MAC Bench finds frontier agents coordinate poorly: 8-agent teams win by exploration, not teamwork
AlexGDimakis · x · 2026-11-19
Alex Dimakis and Bespoke Labs released Multi Agent Coordination Bench: 12 natively multi-agent long-horizon tasks measuring coordination, collaboration, and communication. Key finding: current frontier agents coordinate poorly — a team of 8 beats a single agent, but the gain comes from parallel random exploration, not coordination; of Opus 5.5, GPT-6.1 Sol, and Gemini 4 Argon, only Opus shows signs of gains from coordinating. Tasks (Mars mission, zombie defense, growing a startup) are categorized by Shared vs Distributed information/actions and Collaborative vs Strategic goals; Commons/Strategic variants are excluded from V1.
More from coding & agent
- 130,000+ SEC Form 144 filings added to a dataset, revealing insider sales up to 4 days before Form 4 — virattt · 2026-11-19
- RadixArk trains DaisyRose-27B in 18 days on 144 H200s, doubling Terminal-Bench score — BanghuaZ · 2026-11-19
- Six deployment strategies compared: rolling, blue-green, canary and more — JeremyCMorgan · 2026-11-19
- Dev Argues AI Needs One Interface: Claude or GPT Plus Filesystem or Sandbox — _AustinCalvert_ · 2026-11-19
- 19 Core Concepts of AI System Design: From KV Cache to Multi-Agent Systems — blaizedsouza · 2026-11-19
- 10 Free Repos That Teach AI Agent Harness Engineering: Tools, Memory, Rules — blaizedsouza · 2026-11-19