Frontier AI models remain distant from human designers on design creativity. A new benchmark called Design Creativity Bench evaluates AI models' creativity on UI design tasks. Despite rapid advances in their ability to generate functional websites and product UIs, leading models still struggle to make creative design decisions.
Benchmark scope and design
The benchmark evaluated sixteen frontier models across eighty open-ended UI tasks. Each task was adapted to two distinct product domains and compared against human designer references. The evaluation moves beyond single metrics and measures three dimensions of design creativity:
- Originality: How distinct a model's designs are from other models' outputs for the same brief.
- Creative Range: How much a model adapts its design decisions when the product domain changes.
- Appropriateness: How well the generated UI satisfies the functional requirements of the brief.
Results and gap
The results reveal a substantial gap in overall creativity. The best model, OpenAI GPT-6 Astra, scores 65.5 against a human reference of 92.3. The dataset, leaderboard, and reproducible evaluation framework are being published to make design creativity a measurable dimension of frontier model capability.
Related resources
A paper describing the methodology is available at arxiv.org/abs/2610.11539. The HuggingFace dataset repository hosts the evaluation data at huggingface.co/datasets/Kombai/design-creativity-bench. The source code and benchmark definitions live at github.com/kombai-io/design-creativity-bench.