This research evaluates how effectively different Large Language Models can automate complex Computational Fluid Dynamics (CFD) tasks using OpenFOAM.

• Commercial LLMs demonstrated variable effectiveness in managing CFD tasks like adjusting boundary conditions and solver configurations
• Smaller locally-deployed models (e.g., QwQ-32B) struggled with complex simulation processes
• Zero-shot prompting consistently failed with intricate settings, even for the largest models
• Token costs and performance stability varied significantly across LLM options

This research provides crucial insights for engineering teams seeking cost-effective CFD automation solutions, highlighting the potential and limitations of current LLM technology for specialized engineering applications.

A Status Quo Investigation of Large Language Models towards Cost-Effective CFD Automation with OpenFOAMGPT: ChatGPT vs. Qwen vs. Deepseek