T2MM: An LLM Supported Architecture For Inquiry-Based Modeling

The article introduces the Text to Multimodal Model (T2MM), an architecture designed to enhance inquiry-based modeling in educational contexts by integrating Large Language Models (LLMs) with interactive multimodal capabilities. T2MM facilitates the construction of dynamic models within the Virtual Experimental Research Assistant (VERA) software, allowing for real-time adjustments based on learner input, and outperforms existing LLM-supported code generation approaches in benchmark evaluations. This advancement is significant for practitioners as it demonstrates a pathway for developing more interactive and responsive LLM tools that can better support educational modeling tasks.