Design Creativity Bench Appropriateness
Code · 2026-10-08
Share of each Design Creativity Bench brief's yes/no acceptance criteria met by the model's design, judged per design across the 160 prompts. It is the family's primary because it is the benchmark's only bounded, panel-independent measure.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5.5 | 99.2 |
| GPT-6 Astra | 99.1 |
| GPT-6.1 Sol | 98.8 |
| Claude Fable 5.1 | 98.5 |
| GPT-5.6 Sol | 98.2 |
| Grok 4.7 | 97.8 |
| Muse Spark 1.3 | 97.8 |
| Gemini 3.8 Flash | 97.5 |
| Kimi K3 | 96.8 |
| DeepSeek V4.1 Flash | 96.3 |
| GLM-5.3 Flash | 96.1 |
| GLM-5.3 | 95.6 |
| Grok 4.6 | 95.5 |
| MiniMax M3 | 93.4 |