Pressure Reveals Character - Scheming: Multi-Agent Coordination: leaderboard
Mean 1-5 judge rubric score over 11 multi-turn scenarios testing behaving appropriately in multi-agent settings. One behaviour of the Scheming category; a subset of the headline's transcripts.
Metric: Mean judge rubric score over 11 scenarios (1 = failure, 5 = exemplary). Source: storage.googleapis.com. 24 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.5 | 5 | #250 |
| 2 | Claude Opus 4.5 | 5 | #156 |
| 3 | Claude Sonnet 4 | 4.91 | #400 |
| 4 | GPT-5.2 | 4.82 | #176 |
| 5 | Qwen 3 235B A22B | 4.73 | #623 |
| 6 | Gemini 3 Pro | 4.36 | #137 |
| 7 | GPT-OSS-120B | 4.09 | #700 |
| 8 | Gemini 3 Flash | 4.09 | #164 |
| 9 | GPT-5 Mini | 4.09 | #348 |
| 10 | Kimi K2 | 4.09 | #439 |
| 11 | DeepSeek V3.2 | 3.73 | #406 |
| 12 | Qwen 3 Next 80B A3B | 3.64 | #757 |
| 13 | Qwen 3 14B | 3.45 | #1233 |
| 14 | Gemini 2.5 Pro | 3.36 | #266 |
| 15 | Llama 4 Maverick | 3.18 | #1099 |
Interactive version: theaggregate.ai/benchmark?slug=pressure-reveals-character-scheming-multi-agent-coordination · How It Works · Data refreshed daily, snapshot 2026-11-13.