KIDBench (No Cue) - Content Safety: leaderboard

KIDBench No Cue Content Safety rates whether an LLM answer to a child aged 7 to 11 avoids harmful, explicit or misleading content, when the prompts give no sign that the user is a child (KIDBench, 2026). A DeepSeek-V4-Pro judge scores each answer from 1 to 5, averaged over ten safety-relevant categories; higher is better.

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, no-cue prompts (no sign that the user is a child); higher is better. Source: arxiv.org. Saturation forecast: Around March 2027. 13 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Qwen 3.6 27B4.16#351
2DeepSeek V4 Flash4.15#267
3GPT-5 Mini3.98#364
4Qwen 3.5 4B3.97#1241
5Llama 3.3 70B Instruct3.92#1385
6Gemma 4 31B (IT)3.92#350
7Qwen 3 8B3.9#1873
8Claude Haiku 4.53.87#583
9Gemini 3.1 Flash Lite3.86#371
10Llama 3.1 8B Instruct3.85#4600
11Gemma 3 12B3.75#2059
12KIDBench Llama-3.2-3B (checkpoint unspecified)3.65
13Gemma 3 4B3.55#5339

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=kidbench-no-cue-content-safety · How It Works · Data refreshed daily, snapshot 2026-11-21.