Multi-turn Debate (Lechmazur): leaderboard
Adversarial multi-turn debate benchmark where LLMs argue opposing sides of policy propositions. Each matchup runs twice with sides reversed. Ranked via Bradley-Terry ratings across 1,000+ debates judged by LLMs. Tests factual accuracy under pressure, counterargument quality, and epistemic discipline.
Metric: Bradley-Terry Rating. Source: github.com. Status: years away from saturation. 46 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 5 (High) | 1748.5 |
| 2 | Claude Fable 5 (High) | 1747 |
| 3 | Kimi K3 | 1723.1 |
| 4 | Claude Opus 4.7 (High) | 1672.6 |
| 5 | Muse Spark 1.1 (High) | 1668 |
| 6 | GPT-5.6 Sol (High) | 1664 |
| 7 | Claude Opus 4.8 (High) | 1655.1 |
| 8 | Grok 4.6 (High) | 1628.4 |
| 9 | Claude Sonnet 5 (High) | 1610.8 |
| 10 | GLM-5.2 (Max) | 1585.2 |
| 11 | Claude Sonnet 4.6 (High) | 1585.2 |
| 12 | Qwen 3.8 Max | 1579.4 |
| 13 | GPT-5.4 (High) | 1571.7 |
| 14 | GPT-5.5 (High) | 1551.7 |
| 15 | GLM-5.1 | 1542.8 |
Interactive version: theaggregate.ai/benchmark?slug=multi-turn-debate-lechmazur · How It Works · Data refreshed daily, snapshot 2026-09-05.