XCR-Bench - CSI Prediction: leaderboard
Cross-cultural reasoning from XCR-Bench (Kabir et al., 2026): each culture-specific item in an English sentence is masked, and the model supplies the culturally appropriate Western term. The score counts case-insensitive exact matches per sentence, averaged over sentences and multiplied by 100 (a sentence can hold more than one item); higher is better.
Metric: Exact Match (x100; case-insensitive exact matches of the masked culture-specific items, summed per sentence and averaged over sentences). Source: arxiv.org. Saturation forecast: Estimated already saturated. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude 3.7 Sonnet | 46.76 | #455 |
| 2 | GPT-4o | 43.7 | #741 |
| 3 | Qwen 2.5 Max | 43.7 | #962 |
| 4 | Gemini 2.0 Flash | 42.94 | #702 |
| 5 | Llama 3.3 70B Instruct | 34.54 | #1353 |
| 6 | OLMo 2 32B | 31.87 | #5193 |
| 7 | DeepSeek R1 Distill Qwen 32B | 29.96 | #1780 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=xcr-bench-csi-prediction · How It Works · Data refreshed daily, snapshot 2026-11-12.