ECNU-ChemGPT: A Large Language Model for Chemistry and Retrosynthesis Predictions

Abstract

Large language models (LLMs) have achieved impressive progress across a broad range of general-purpose tasks, but their effectiveness in chemistry remains limited due to scarce domain-specific datasets, and the demand for precise symbolic and structural reasoning. Here we introduce ECNU-ChemGPT (East China Normal University Chemistry GPT model), a chemistry-specialized LLM engineered for deep chemical knowledge understanding and accurate retrosynthetic route planning. Our approach is distinguished by four key strategies: structured prompt-based knowledge distillation from authoritative chemistry textbooks to construct a high-quality question-answering dataset; domain-specific prompt engineering using curated chemical keywords, combined with LLM APIs for data derivation and knowledge distillation; fine-tuning on a meticulously cleaned Pistachio reaction dataset to enhance retrosynthesis prediction accuracy; and integration of BrainGPT (a brain-inspired multi-model scheduling framework), a dynamic multi-model scheduling framework that enables task-specific invocation of multiple specialized models trained for diverse chemistryrelated tasks. ECNU-ChemGPT exhibits superior performance on chemistry questionanswering and retrosynthetic planning benchmarks, outperforming leading generalpurpose models-including DeepSeek-R1, Qwen-2.5, and GPT-4o. In retrosynthesis, it achieves a Top-1 accuracy of 68.3% on the USPTO-50K dataset and successfully reproduces seven complete drug synthesis routes reported in the literatures and patents. These results underscore the effectiveness of domain-adapted LLM fine-tuning combined with dynamic multi-model task scheduling, providing a scalable and robust solution for chemical knowledge question answering, and retrosynthetic planning.