ACL Findings2026
TOOLCAD: Exploring Tool-Using Large Language Models in Text-to-CAD Generation with Reinforcement Learning
Xing Wu Yifei Gong, Tukang Wenda Liu
Abstract
Computer-Aided Design (CAD) is an expertlevel task that relies on long-horizon reasoning and coherent modeling actions. Large Language Models (LLMs) have shown remarkable advancements in enabling language agents to tackle real-world tasks. Notably, there has been no investigation into how tool-using LLMs optimally interact with CAD engines, hindering the emergence of LLM-based agentic text-to-CAD modeling systems. We propose TOOL-CAD, a novel agentic CAD framework deploying LLMs as tool-using agents for text-to-CAD generation. Furthermore, we introduce an interactive CAD modeling gym to rollout reasoning and tool-augmented interaction trajectories with the CAD engine, incorporating hybrid feedback and human supervision. Meanwhile, an end-to-end post-training strategy is presented to enable LLMs to elicit refined CAD Modeling Chain of Thought (CAD-CoT) and evolve into proficient CAD tool-using agents via curriculum online reinforcement learning. Our findings demonstrate TOOLCAD fills the gap in adopting and training open-source LLMs for CAD tool-using agents, enabling them to perform comparably to proprietary models, paving the way for more accessible and robust autonomous text-to-CAD modeling systems. 1
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
- Text2CAD: Generating Sequential CAD Designs from Beginner-to-Expert Level Text Prompts cites Hierarchical Neural Coding for Controllable CAD Model Generation
- Text2CAD: Generating Sequential CAD Designs from Beginner-to-Expert Level Text Prompts cites SkexGen: Autoregressive Generation of CAD Construction Sequences with Disentangled Codebooks
- ToolRL: Reward is All Tool Learning Needs cites Direct Preference Optimization: Your Language Model is Secretly a Reward Model
Builds on13
- Direct Preference Optimization: Your Language Model is Secretly a Reward ModelRafael Rafailov, Archit Sharma, Eric Mitchell, Christopher D. Manning et al.NeurIPS 2023 · 10,924 citations
- ToolRL: Reward is All Tool Learning NeedsCheng Qian, Emre Can Acikgoz, Qi He, Hongru Wang et al.NeurIPS 2025 · 387 citations
- Text2CAD: Generating Sequential CAD Designs from Beginner-to-Expert Level Text PromptsMohammad Sadil Khan, Sankalp Sinha, Talha Uddin Sheikh, Didier Stricker et al.NeurIPS 2024 · 148 citations
- SkexGen: Autoregressive Generation of CAD Construction Sequences with Disentangled CodebooksXiang Xu, Karl D. D. Willis, Joseph G. Lambourne, Chin-Yi Cheng et al.ICML 2022 · 126 citations
- Hierarchical Neural Coding for Controllable CAD Model GenerationXiang Xu, Pradeep Kumar Jayaraman, Joseph George Lambourne, Karl D. D. Willis et al.ICML 2023 · 88 citations
Related papers
- From Intent to Execution: Multimodal Chain-of-Thought Reinforcement Learning for Precise CAD Code GenerationKe Niu, Haiyang Yu, Zhuofan Chen, Mengyang Zhao et al.AAAI 2026 · 5 citations
- CAD-Coder: Text-to-CAD Generation with Chain-of-Thought and Geometric RewardYandong Guan, Xilin Wang, Ximing Xing, Jing Zhang et al.NeurIPS 2025 · 64 citations
- Multi-Agent CAD Code GenerationYang Liu, Daxuan Ren, Yijie Ding, Jianmin Zheng et al.SIGGRAPH 2026
- Automated CAD Modeling Sequence Generation from Text Descriptions via Transformer-Based Large Language ModelsJianxing Liao, Junyan Xu, Yatao Sun, Maowen Tang et al.ACL 2025 · 8 citations
- Towards High-Fidelity CAD Generation via LLM-Driven Program Generation and Text-Based B-Rep Primitive GroundingJiahao Li, Qingwang Zhang, Qiuyu Chen, Guozhan Qiu et al.ICML 2026 · 7 citations
