KIDBench (Implicit Cue) - Content Safety: leaderboard
Metric: Content Safety score (1-5): whether the response avoids harmful, explicit or misleading content, rated 1-5 by a DeepSeek-V4-Pro judge with the KIDBench rubric and category-specific rules, temperature 0 for responder and judge, averaged 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: Estimated already saturated. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.6 27B | 4.68 |
| 2 | Gemini 3.1 Flash Lite | 4.57 |
| 3 | DeepSeek V4 Flash | 4.55 |
| 4 | Gemma 4 31B (IT) | 4.49 |
| 5 | Claude Haiku 4.5 | 4.37 |
| 6 | GPT-5 Mini | 4.35 |
| 7 | Qwen 3.5 4B | 4.31 |
| 8 | Gemma 3 12B | 4.28 |
| 9 | Llama 3.3 70B Instruct | 4.25 |
| 10 | Qwen 3 8B | 3.99 |
| 11 | Llama 3.1 8B Instruct | 3.86 |
| 12 | Gemma 3 4B | 3.79 |
Interactive version: theaggregate.ai/benchmark?slug=kidbench-implicit-cue-content-safety · How It Works · Data refreshed daily, snapshot 2026-10-07.