Awesome AI4CAD Hub Papers · Datasets · Projects
← Back to papers

Text2CAD-Bench: A Benchmark for LLM-based Text-to-Parametric CAD Generation

arXiv 2026 48.3 method

TLDR

Introduces Text2CAD-Bench, a benchmark for evaluating text-to-parametric CAD generation across geometric complexity and real-world domains.

Reasoning

The paper's strength lies in its systematic benchmark design with 600 human-curated examples spanning four complexity levels and dual-style prompts, addressing a gap in existing benchmarks. Weaknesses include reliance on evaluating existing models without proposing a new method, and limited detail on the benchmark's construction or validation in the abstract.

Read-first score

Read-first score 48.3, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 67.

Recency 8%
100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Methodology quality 25%
60

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

Topical relevance 42%
41.9

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%
30

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 39.

Keyword Scores

CAD generation
10
text-to-CAD
10
parametric CAD
9
generative CAD
8
AI for CAD
7
computer-aided design
6
sketch extrusion
5
neural CAD
2
CAD reconstruction
2
CAD program
2
B-Rep
1
boundary representation
1
constructive solid geometry
1
CSG
1
image-to-CAD
1
point cloud to CAD
1

Deep Analysis

Innovations

  • First benchmark systematically evaluating text-to-CAD across geometric complexity and application diversity
  • Hierarchical benchmark with four levels (L1-L4) from basic primitives to real-world domains beyond mechanical parts
  • Dual-style prompts: geometric descriptions for non-expert users and procedural sequences for expert-level conventions

Methodology

We construct Text2CAD-Bench, a benchmark of 600 human-curated examples spanning four levels of geometric complexity and application diversity, each paired with geometric and procedural prompts. We evaluate mainstream general LLMs and domain-specific models on this benchmark.

Key Results

Current models perform reasonably on basic geometry but degrade substantially on complex topology and advanced features.

Tags

LG