Text-to-CadQuery: A New Paradigm for CAD Generation with Scalable Large Model Capabilities

arXiv 2025 55.6 method

TLDR

Proposes generating CadQuery code from text using fine-tuned LLMs, achieving significant improvements in exact match and Chamfer Distance.

Reasoning

Strengths include a novel paradigm leveraging pretrained LLMs for direct CAD code generation, with strong empirical results across model scales. Weaknesses are the reliance on CadQuery-specific representation and limited evaluation metrics that may not capture full practical usability.

Read-first score

Read-first score 55.6, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 80.

Recency 8%

86.7

Uses a gentle age decay so recent papers surface without erasing older foundations. 2025

Methodology quality 25%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=dataset,metric

Topical relevance 42%

Uses existing LLM keyword relevance scores normalized to 0-100. AI for CAD,computer-aided design,neural CAD,generative CAD,parametric CAD,B-Rep,boundary representation,constructive solid geometry,CSG,sketch extrusion,CAD generation,CAD reconstruction,text-to-CAD,image-to-CAD,point cloud to CAD,CAD program

Reproducibility 25%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=code,dataset,github

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 17.

Keyword Scores

text-to-CAD

computer-aided design

generative CAD

CAD generation

CAD program

AI for CAD

parametric CAD

neural CAD

B-Rep

boundary representation

CAD reconstruction

constructive solid geometry

CSG

sketch extrusion

image-to-CAD

point cloud to CAD

Deep Analysis

Innovations

Direct generation of CadQuery code from text, eliminating intermediate task-specific command sequences
Leveraging pretrained LLMs' Python generation and spatial reasoning for CAD generation
Augmenting the Text2CAD dataset with 170,000 CadQuery annotations
Demonstrating that larger models yield better performance after fine-tuning on Text-to-CadQuery

Methodology

Fine-tune six open-source LLMs of varying sizes on a dataset of 170,000 text-to-CadQuery pairs derived from Text2CAD. Models generate CadQuery Python code directly from natural language, evaluated via top-1 exact match and Chamfer Distance.

Key Results

The best model achieves a top-1 exact match of 69.3% (up from 58.8%) and reduces Chamfer Distance by 48.6%.

Resources

Paper