CoCoBench: leaderboard
Metric: Task success rate (%; share of episodes whose goal predicates all hold at the end; 897 oracle-validated executable household multi-agent task instances with two to four agents, centralized planner with image observations, task-specific step budgets). Source: arxiv.org. Saturation forecast: Around December 2026. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.6 Sol | 84.8 |
| 2 | Claude Opus 4.8 | 78.5 |
| 3 | Qwen 3.6 Plus | 69.1 |
| 4 | Claude Haiku 4.5 | 63.1 |
| 5 | Kimi K2.6 | 57.6 |
| 6 | Qwen 3.5 9B | 36.7 |
| 7 | GPT-5.4 Mini | 33.6 |
| 8 | Qwen 3 VL 8B Instruct | 24.6 |
Interactive version: theaggregate.ai/benchmark?slug=cocobench · How It Works · Data refreshed daily, snapshot 2026-09-26.