BanglaSocialBench - Kinship Reasoning: leaderboard
Metric: Accuracy (%) on the 345 four-option Bengali kinship puzzles (multi-hop lineage descriptions; pick the correct kinship term), zero-shot Bangla prompts at temperature 0, culturally grounded scenarios written and verified by native Bangla speakers; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Flash | 92.46 | #237 |
| 2 | Claude Sonnet 4 | 88.98 | #194 |
| 3 | Gemini 2.0 Flash | 68.12 | #331 |
| 4 | GPT-4o | 67.83 | #333 |
| 5 | DeepSeek V3.1 | 64.06 | #260 |
| 6 | Llama 3.3 70B Instruct | 51.01 | #520 |
| 7 | GPT-4o Mini | 49.86 | #588 |
| 8 | Qwen 2.5 72B Instruct | 48.7 | #436 |
| 9 | Gemma 3 27B | 48.12 | #596 |
| 10 | Claude 3.5 Haiku | 44.06 | #553 |
| 11 | Gemma 3 12B | 42.61 | #666 |
| 12 | Llama 3 8B Instruct | 31.88 | #1115 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=banglasocialbench-kinship-reasoning · How It Works · Data refreshed daily, snapshot 2026-10-11.