BrickBench: A Benchmark for Agentic Text-Conditioned LEGO Set Design

Forum topic · 小凯 · 2026-10-10

Summary

BrickBench (arXiv:2610.12452) is a benchmark introduced by researchers including Peter Kulits, Yiqing Xu, R. Kenny Jones, Cordelia Schmid, and Jiajun Wu for evaluating agentic text-conditioned LEGO set design. Given a text prompt, an agent must produce an assembly that satisfies semantic and design criteria while remaining physically buildable: it must select parts from a discrete library and reason jointly about local and global constraints. The benchmark scores validity, alignment, and design quality across three settings that vary in scale and part availability. The authors also release BrickAgent, an environment that lets coding agents construct, inspect, and validate their designs. Findings show that leading AI agents largely satisfy verifiable physical and semantic requirements but still fall short of human-designed sets. The benchmark and environment are open-sourced at http://www.brickben.ch.

Paper Overview

Research Area: CV Authors: Peter Kulits, Yiqing Xu, R. Kenny Jones, Cordelia Schmid, Jiajun Wu Released: 2026-10-08 arXiv: 2610.12452

Abstract

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. To do so, it must select parts from a discrete library and reason jointly about local and global constraints. We score validity, alignment, and design across three settings that vary in scale and part availability. We provide BrickAgent, an environment for coding agents to construct, inspect, and validate their designs. We find that leading agents largely satisfy verifiable physical and semantic requirements, but fall short of human designs. We release our benchmark and environment at http://www.brickben.ch

Key Points

Task: agents must turn a text prompt into a LEGO assembly that is semantically aligned, well-designed, and physically buildable.
Constraints: parts must be selected from a discrete library; agents must reason jointly about local and global constraints.
Evaluation: validity, alignment, and design quality scored across three settings varying in scale and part availability.
Tooling: BrickAgent environment lets coding agents construct, inspect, and validate designs.
Findings: leading agents largely satisfy verifiable physical and semantic requirements, but fall short of human designs.
Release: benchmark and environment available at http://www.brickben.ch

--- *Auto-collected on 2026-10-10*