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

UniCAD: A Unified Benchmark and Universal Model for Multi-Modal Multi-Task CAD

arXiv 2026 69 method

TLDR

UniCAD introduces a unified benchmark and a multi-modal large language model for diverse CAD tasks, achieving state-of-the-art results.

Reasoning

The paper's strength lies in its comprehensive benchmark covering multiple CAD tasks and modalities, along with a universal model that outperforms baselines. However, the abstract lacks details on methodology and limitations, and some keywords like B-Rep and CSG are not addressed.

Read-first score

Read-first score 69, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 93.

Methodology quality 25%
100

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

Recency 8%
100

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

Topical relevance 42%
58.1

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%
46

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

Field roles

FrontierMethodology anchor

Rank sensitivity

Stability: volatile; rank range: 5.

Keyword Scores

computer-aided design
10
text-to-CAD
10
image-to-CAD
10
point cloud to CAD
10
AI for CAD
9
generative CAD
9
CAD generation
9
CAD reconstruction
9
neural CAD
8
parametric CAD
5
sketch extrusion
4
B-Rep
0
boundary representation
0
constructive solid geometry
0
CSG
0
CAD program
0

Deep Analysis

Innovations

  • UniCAD benchmark: a unified multi-modal multi-task CAD benchmark covering point-to-CAD reconstruction, text/image-to-CAD generation, and CAD question answering.
  • UniCAD-MLLM: a universal multi-modal large language model that processes text, images, sketches, and point clouds to perform heterogeneous CAD tasks end-to-end in a single framework.
  • State-of-the-art performance across all tasks on UniCAD and Fusion360, surpassing task-specific and multi-task baselines.

Methodology

UniCAD-MLLM is a multi-modal large language model that ingests text, images, sketches, and point clouds and performs point-to-CAD reconstruction, text/image-to-CAD generation, and CAD question answering in an end-to-end manner within a single framework. The UniCAD benchmark provides a unified evaluation suite for these tasks across diverse input modalities.

Key Results

UniCAD-MLLM achieves state-of-the-art results on all tasks in the UniCAD and Fusion360 benchmarks, outperforming both task-specific and multi-task baselines.

Tags

CVAI