Foundation Models for Automatic CAD Generation
Quick Answer
The study introduces LLMForge, a multi-model framework for automatic CAD generation, achieving 98.97% mesh success with models like DeepSeek-V3.2 and Qwen3-235B-A22B.
Quick Take
The study introduces LLMForge, a multi-model framework for automatic CAD generation, achieving 98.97% mesh success with models like DeepSeek-V3.2 and Qwen3-235B-A22B. It highlights the effectiveness of compact instruction-tuned models compared to larger systems, while also addressing challenges in generating rotationally symmetric geometries.
Key Points
- LLMForge integrates JSON-schema validation and multi-round iterative refinement for CAD generation.
- IterTracer achieved 98.97% mesh success using analytic visual metrics for feedback.
- -based critique in IterVision enabled 100% watertight mesh generation on leading models.
- Seven foundation models were evaluated across 97 engineering design problems.
- Challenges remain in generating geometries like cylinders due to scoring divergence.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Recent advances in Large Language Models (LLMs) and Vision-Language Models (VLMs) enable the automatic generation of parametric 3D designs from natural-language specifications. This chapter presents an empirical study of foundation models for automatic Computer-Aided Design (CAD) generation of mechanical parts, using a unified evaluation pipeline and a curated benchmark of 97 engineering design problems. We introduce LLMForge, a multi-model text-to-CAD framework integrating JSON-schema validation, analytic feature scoring, mesh synthesis, and multi-round iterative refinement, studied under two critique regimes. IterTracer uses a Phong-shaded ray-trace renderer with analytic visual metrics (silhouette IoU, hole visibility, edge clearance, aspect-ratio conformance) for lightweight geometry-aware feedback across rounds. IterVision replaces the analytic scorer with a VLM semantic critic (Qwen2.5-VL-72B) that evaluates rendered views via chain-of-thought visual reasoning, assessing spatial coherence and design intent. On a benchmark spanning four canonical geometry families (plates with holes and bolt circles, multi-feature boxes, flanged cylinders, and L-brackets), we evaluate seven foundation models: DeepSeek-V3.2, Qwen3-235B-A22B, Llama-3.3-70B, Gemma-3-27B, GLM-4.5, MiniMax-M2.1, and INTELLECT. Under IterTracer, the four highest-ranked models form a tight cluster (overall mean in [0.885, 0.890]) with 98.97% mesh success, showing that compact instruction-tuned models can match substantially larger systems. VLM-based critique in IterVision yields 100% watertight mesh generation on the leading model while surfacing systematic difficulty on rotationally symmetric geometries such as cylinders, where visual and semantic scoring diverge most. We discuss benchmark design, failure modes, CAD-oriented prompting, and implications for industrial workflows and scalable automated mechanical design.
| Comments: | Accepted as a book chapter in "Advances in Global Applied Artificial Intelligence" (G. A. Tsihrintzis, M. Virvou, N. G. Bourbakis, L. C. Jain, Eds.), authenticated version will be published in Springer series: Learning and Analytics in Intelligent Systems |
| Subjects: | Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE) |
| Cite as: | arXiv:2607.05573 [cs.AI] |
| (or arXiv:2607.05573v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2607.05573 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: J. De Curtò [view email]
[v1]
Mon, 6 Jul 2026 19:17:49 UTC (1,375 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.AI
See more →Onnes: A Physics-Grounded LLM Simulator for Cryogenic Fault Diagnosis in Quantum Computing Infrastructure
Onnes is a physics-grounded digital twin simulator for dilution refrigerators, enhancing cryogenic fault diagnosis in quantum computing. It achieves a classification accuracy of 99.0% using few-shot demonstrations, matching a supervised ML classifier, while maintaining a low false alarm rate of 6.4% on real hardware.