KIDBench (Implicit Cue): leaderboard
Metric: Total score (1-5): mean of the five child-safety dimensions (content safety, developmental appropriateness, socio-emotional support, moral guidance, boundary-setting), each rated 1-5 by a DeepSeek-V4-Pro judge with the KIDBench rubric and category-specific rules, temperature 0 for responder and judge, on the KIDBench child-safety prompts for children aged 7 to 11 (50 per category, 10 categories), single-turn, implicit-cue prompts (wording or situation suggests a child speaker, age not stated); higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V4 Flash | 4.43 |
| 2 | Qwen 3.6 27B | 4.29 |
| 3 | Gemini 3.1 Flash Lite | 4.27 |
| 4 | Gemma 4 31B (IT) | 4.14 |
| 5 | Claude Haiku 4.5 | 4 |
| 6 | GPT-5 Mini | 3.99 |
| 7 | Gemma 3 12B | 3.98 |
| 8 | Qwen 3.5 4B | 3.9 |
| 9 | Llama 3.3 70B Instruct | 3.76 |
| 10 | Qwen 3 8B | 3.59 |
| 11 | Gemma 3 4B | 3.52 |
| 12 | Llama 3.1 8B Instruct | 3.11 |
Interactive version: theaggregate.ai/benchmark?slug=kidbench-implicit-cue · How It Works · Data refreshed daily, snapshot 2026-10-07.