Alibaba / Qwen3.6 27B
Composite score—
Model results
Ranked by composite score across assistant and child-aware prompt variants.
Anthropic / Claude Opus 4.8
Composite score—
Anthropic / Claude Sonnet 4.6
Composite score—
Anthropic / Claude Haiku 4.5
Composite score—
DeepSeek / DeepSeek V4 Pro
Composite score—
Google / Gemini 3.1 Pro
Composite score—
Google / Gemini 3.5 Flash
Composite score—
Google / Nano Banana 2
Composite score—
Meta / Llama 4 Scout
Composite score—
Meta / Llama 4 Maverick
Composite score—
MiniMax / MiniMax M2.5
Composite score—
Mistral / Mistral Large 3
Composite score—
Mistral / Mistral Medium 3.5
Composite score—
Mistral / Mistral Small 4
Composite score—
Moonshot AI / Kimi K2.6
Composite score—
Nvidia / Nemotron 3 Ultra
Composite score—
OpenAI / GPT 5.5
Composite score—
xAI / Grok 4.3
Composite score—
Xiaomi / MiMo V2.5 Pro
Composite score—
Z.ai / GLM-5V Turbo
Composite score—