LLM Emergent Collusion: leaderboard
Measures the rate at which frontier LLMs spontaneously form price-fixing cartels in a sealed-bid auction simulation: without being instructed to collude. Tests emergent strategic coordination and anti-competitive behavior.
Metric: Collusion Rate (%). Source: github.com. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Grok 3 | 19 | #304 |
| 2 | GPT-4o (Mar 2025) | 23 | #330 |
| 3 | DeepSeek V3 (0324) | 26 | #340 |
| 4 | Claude Sonnet 4 (Non-reasoning) | 32 | #199 (Claude Sonnet 4) |
| 5 | Claude Opus 4 (Non-reasoning) | 36 | #163 (Claude Opus 4) |
| 6 | Gemini 2.5 Pro | 39 | #152 |
| 7 | Mistral Medium 3 | 47 | #500 |
| 8 | O3 | 51 | #124 |
| 9 | Qwen 3 235B A22B | 57 | #313 |
| 10 | O4 Mini | 62 | #180 |
| 11 | DeepSeek R1 0528 | 71 | #222 |
| 12 | Grok 4 | 75 | #181 |
Interactive version: theaggregate.ai/benchmark?slug=llm-emergent-collusion · How It Works · Data refreshed daily, snapshot 2026-10-20.