Text-to-CAD Evaluation with CADTests

2026-05-08

AI Summary

Existing text-to-CAD generation models lack effective evaluation protocols, making it difficult to verify whether generated outputs satisfy the geometric and topological constraints specified in textual prompts. This work proposes a novel evaluation paradigm based on automated executable tests and introduces CADTestBench, the first test-driven benchmark for this task. By employing procedural CADTests, the framework enables systematic assessment of state-of-the-art methods. Furthermore, the study innovatively integrates test feedback directly into the generation process, yielding a simple yet effective test-guided baseline model that outperforms current approaches. These results demonstrate the efficacy of test signals in enhancing the fidelity and correctness of generated CAD models.

Abstract

Text-to-CAD has recently emerged as an important task with the potential to substantially accelerate design workflows. Despite its significance, there has been surprisingly little work on Text-to-CAD evaluation, and assessing CAD model generation performance remains a considerable challenge. In this work, we introduce a new evaluation perspective for Text-to-CAD based on automated testing. We propose CADTestBench, the first test-based benchmark for Text-to-CAD, based on CADTests, executable software tests that verify whether a generated CAD model satisfies the geometric and topological requirements of the input prompt. Using CADTestBench, we conduct comprehensive benchmarking of recent Text-to-CAD methods and further demonstrate that CADTests can also guide CAD model generation, yielding simple baselines that surpass performance of current methods. CADTestBench code and data are available at GitHub and Hugging Face dataset.

Problem

• Text-to-CAD
• evaluation
• CAD model generation
• geometric requirements
• topological requirements

Innovation

• Text-to-CAD
• CADTests
• automated evaluation
• CADTestBench
• generative design