KIDBench - Cultural Alignment (India): leaderboard
Metric: Cultural Alignment score (1-5): whether the response adapts guidance, help-seeking pathways and social expectations to India, rated by a DeepSeek-V4-Pro judge with country-specific cultural rules (temperature 0), on the 500 implicit-cue KIDBench prompts in English with a system instruction naming a child aged 7 to 11 in India, mean over the 10 categories, single-turn; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.6 27B | 4.41 |
| 2 | Gemini 3.1 Flash Lite | 3.84 |
| 3 | Gemma 4 31B (IT) | 3.83 |
| 4 | Gemma 3 12B | 3.78 |
| 5 | DeepSeek V4 Flash | 3.68 |
| 6 | GPT-5 Mini | 3.64 |
| 7 | Claude Haiku 4.5 | 3.4 |
| 8 | Qwen 3.5 4B | 3.29 |
| 9 | Gemma 3 4B | 3.27 |
| 10 | Qwen 3 8B | 3.12 |
| 11 | Llama 3.3 70B Instruct | 3.08 |
| 12 | Llama 3.1 8B Instruct | 2.75 |
Interactive version: theaggregate.ai/benchmark?slug=kidbench-cultural-alignment-india · How It Works · Data refreshed daily, snapshot 2026-10-07.