MAC Bench - Single Agent: leaderboard

Multi Agent Coordination Bench's baseline: one agent alone works each of the twelve four-hour team tasks for the same wall-clock time (scored 0-100 by a hidden deterministic grader), the reference the team-of-eight board is compared against.

Metric: Mean final score x100, one agent alone. Source: benchmarks.bespokelabs.ai. Saturation forecast: Around February 2028. 3 models tracked. Updated 19 Nov 2026.

Top models

#ModelScoreOverall rank
1Claude Opus 5.538.71#9
2GPT-6.1 Sol38.2#15
3Gemini 4 Argon28#18

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=mac-bench-single-agent · How It Works · Data refreshed daily, snapshot 2026-11-20.