KIDBench (Implicit Cue) - Hate, Bias, and Identity Attacks: leaderboard

KIDBench Implicit Cue Hate, Bias, and Identity Attacks covers the 50 questions from children aged 7 to 11 in the Hate, Bias, and Identity Attacks category of KIDBench (2026), asked so that the prompt wording or situation suggests a child speaker without stating an age. A DeepSeek-V4-Pro judge rates each answer on five child-safety dimensions from 1 to 5; the board shows their mean, higher is better.

Metric: Category score (1-5): mean of the five child-safety dimensions, each rated 1-5 by a DeepSeek-V4-Pro judge with the KIDBench rubric and category-specific rules, temperature 0 for responder and judge, over the 50 Hate, Bias, and Identity Attacks prompts of KIDBench for children aged 7 to 11, 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. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1DeepSeek V4 Flash4.76#267
2GPT-5 Mini4.6#364
3Gemma 4 31B (IT)4.58#350
4Gemini 3.1 Flash Lite4.57#371
5Gemma 3 12B4.54#2059
6Qwen 3.6 27B4.49#351
7Llama 3.3 70B Instruct4.3#1385
8Qwen 3.5 4B4.18#1241
9Gemma 3 4B4.14#5339
10Qwen 3 8B4.05#1873
11Claude Haiku 4.53.95#583
12Llama 3.1 8B Instruct3.61#4600
13KIDBench Llama-3.2-3B (checkpoint unspecified)3.42

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

Interactive version: theaggregate.ai/benchmark?slug=kidbench-implicit-cue-hate-bias-and-identity-attacks · How It Works · Data refreshed daily, snapshot 2026-11-21.