MiGUE-Bench - Cross-Document Subevent Relations: leaderboard
Metric: Accuracy (%) on 300 cross-document subevent relation multiple-choice questions with LLM-built distractors over news documents; higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V3.2 | 84.8 |
| 2 | Claude Opus 4.5 | 84.21 |
| 3 | Qwen 3 235B A22B | 83.62 |
| 4 | GLM-4.7 | 83 |
| 5 | GPT-5.2 Pro | 81.88 |
| 6 | Qwen 3 Max | 80.99 |
| 7 | Gemini 3 Pro | 80.34 |
| 8 | Qwen 3 30B A3B | 78.95 |
| 9 | Kimi K2 | 78.07 |
| 10 | Claude Haiku 4.5 | 71.64 |
| 11 | Qwen 3 8B | 61.7 |
Interactive version: theaggregate.ai/benchmark?slug=migue-bench-cross-document-subevent-relations · How It Works · Data refreshed daily, snapshot 2026-09-29.