BrickBench: Evaluating Agentic Brick Design
arXiv cs.AI · 9 Oct 2026, 03:58 AEST
BrickBench is a new benchmark for agentic text-conditioned LEGO-set design. It evaluates how well agents can create physically buildable assemblies based on given prompts.
Key facts
- We propose BrickBench, a benchmark for agentic text-conditioned LEGO-set design.
- Given a prompt, an agent is tasked with producing an assembly that not only satisfies semantic and design criteria, but that can also be physically built.
- We score validity, alignment, and design across three settings that vary in scale and part availability.
- We find that leading agents largely satisfy verifiable physical and semantic requirements, but fall short of human designs.
Read aloud
BrickBench is a new benchmark for agentic text-conditioned LEGO-set design. It evaluates how well agents can create physically buildable assemblies based on given prompts. We propose BrickBench, a benchmark for agentic text-conditioned LEGO-set design. Given a prompt, an agent is tasked with producing an assembly that not only satisfies semantic and design criteria, but that can also be physically built. We score validity, alignment, and design across three settings that vary in scale and part availability. We find that leading agents largely satisfy verifiable physical and semantic requirements, but fall short of human designs.
Sources in this cluster
- BrickBench: Evaluating Agentic Brick Design
arXiv cs.AI · 9 Oct 2026, 03:58 AEST