HumRightsBench - Rule Recall: leaderboard
HumRightsBench (2026) is a pilot, expert-validated benchmark of reasoning in international human rights law built on right-to-water scenarios and structured along issue identification, rule recall, rule application and proposed remedies. This board asks which international human rights rule, named by instrument and article, applies to the sub-scenario. Accuracy in percent over five seeds; higher is better.
Metric: Accuracy (%; x100 of the exact-set match rate on the rule-recall multiple-choice question: which international human rights rule applies, rules named by instrument and article; pilot HumRightsBench question set on right-to-water scenarios validated by human rights experts; each question asked with five seeds and shuffled answer order (Qwen 3.5-9B multiple-choice types over four), responses through each provider's structured-output interface, errored calls counted as incorrect). Source: arxiv.org. Saturation forecast: Around February 2027. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.7 | 77.4 | #75 |
| 2 | Gemini 3 Flash (Preview) | 76.5 | #141 |
| 3 | GPT-5 | 71.5 | #163 |
| 4 | Qwen 3.5 9B | 49.4 | #803 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=humrightsbench-rule-recall · How It Works · Data refreshed daily, snapshot 2026-11-12.