LongJudgeBench (Rubric): leaderboard
Metric: Judge accuracy (%) averaged over the six datasets (DeepResearch Bench, RealDeepResearch, SurGE structure and content, WritingPreferenceBench, VerifyBench, MA): agreement of the LLM judge's pairwise preferences (from pointwise scores, pairwise choices or listwise ranks) or correctness labels with the human reference judgments, averaged within query groups, temperature 0, judging with task-specific rubrics; higher is better. Source: arxiv.org. Saturation forecast: Around July 2027. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 Max | 61.6 |
| 2 | GLM-5.1 | 61.47 |
| 3 | Kimi K2.6 | 61.33 |
| 4 | DeepSeek V4 Flash | 58.08 |
| 5 | Qwen 3 32B (Non-reasoning) | 55.94 |
| 6 | GPT-5.2 | 53.18 |
| 7 | Qwen 3 32B (Thinking) | 51.07 |
| 8 | GPT-4o Mini | 43.03 |
Interactive version: theaggregate.ai/benchmark?slug=longjudgebench-rubric · How It Works · Data refreshed daily, snapshot 2026-09-29.