CDH-Bench (Binary QA) - Relational Anomalies: leaderboard
Metric: Counterfactual accuracy (%) on the 100 relational-anomaly images (reversed behavioural, spatial, size or causal relations such as mice hunting a cat), answering a yes/no question that asserts the commonsense reading and denies the depicted anomaly (for example 'Does the mouth have 2 rows of teeth and not 3 rows?'), so the correct answer on every counterfactual image is no; each image is shown alone, without its matched commonsense twin, under one prompt per format; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 90 | #54 |
| 2 | Kimi K2.5 | 80 | #139 |
| 3 | Doubao-Seed-1-8-251228 | 80 | #146 |
| 4 | Gemini 3.1 Flash Lite (Preview) | 79 | #186 |
| 5 | Qwen 3.5 Plus | 74.49 | #123 |
| 6 | GPT-5.4 | 62 | #76 |
| 7 | GPT-5.4 Mini | 62 | #202 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=cdh-bench-binary-qa-relational-anomalies · How It Works · Data refreshed daily, snapshot 2026-10-11.