Agentic reinforcement learning empowers next-generation chemical language models for molecular design and synthesis

Abstract Language models are revolutionizing the biochemistry domain, assisting scientists in drug design and chemical synthesis with high efficiency. Yet current approaches struggle between small language models prone to hallucination and limited knowledge retention, and large cloud-based language models plagued by privacy risks and high inference costs. To bridge this gap, we introduce ChemCRAFT, a novel framework leveraging agentic reinforcement learning to decouple chemical reasoning from knowledge storage. Instead of forcing the model to memorize vast chemical data, our approach empowers the language model to interact with a sandbox for precise information retrieval. This externalization of knowledge allows a locally deployable small model to achieve superior performance while being highly token-efficient compared to traditional multi-agent systems. To enable small language models for agent-calling ability, we build an agentic trajectory construction pipeline and a comprehensive chemical-agent sandbox. Based on sandbox interactions, we constructed ChemToolDataset, the first large-scale chemical tool trajectory dataset. Simultaneously, we propose SMILES-GRPO to build a dense chemical reward function, promoting the model's ability to call chemical agents. Evaluations across diverse aspects of drug design show that ChemCRAFT outperforms current cloud-based LLMs in molecular structure analysis, molecular optimization, and synthesis pathway prediction, demonstrating that scientific reasoning is not solely an emergent ability of model scale, but a learnable policy of tool orchestration. Furthermore, assessments in real-world drug optimization scenarios, including fentanyl analog optimization and pro-inflammatory GSTO1 inhibitors optimization, reveal that ChemCRAFT generates optimization trajectories mirroring human experts, confirming its reliability as a scientific assistant. This work establishes a cost-effective and privacy-preserving paradigm for AI-aided chemistry, opening new avenues for accelerating molecular discovery with locally deployable agents.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-17
DOI
https://doi.org/10.21203/rs.3.rs-9959783/v1
Citations
12
Primary Topic
Computational Drug Discovery Methods
Type
preprint

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Agentic reinforcement learning empowers next-generation chemical language models for molecular design and synthesis

Liuzhenghao Lv, Yonghong Tian, Yu Li, Yang Zhang et al.
12 citations
Research Square
Computational Drug Discovery Methods
preprint

Agentic reinforcement learning empowers next-generation chemical language models for molecular design and synthesis

Liuzhenghao Lv, Yonghong Tian, Yu Li, Yang Zhang, Li Yuan, He Cao, Zhiyuan Yan, Bin Feng, Zijing Liu, Hao Li, Shenyao Peng, Yu Wang
preprint en
12 citations

Abstract

Abstract Language models are revolutionizing the biochemistry domain, assisting scientists in drug design and chemical synthesis with high efficiency. Yet current approaches struggle between small language models prone to hallucination and limited knowledge retention, and large cloud-based language models plagued by privacy risks and high inference costs. To bridge this gap, we introduce ChemCRAFT, a novel framework leveraging agentic reinforcement learning to decouple chemical reasoning from knowledge storage. Instead of forcing the model to memorize vast chemical data, our approach empowers the language model to interact with a sandbox for precise information retrieval. This externalization of knowledge allows a locally deployable small model to achieve superior performance while being highly token-efficient compared to traditional multi-agent systems. To enable small language models for agent-calling ability, we build an agentic trajectory construction pipeline and a comprehensive chemical-agent sandbox. Based on sandbox interactions, we constructed ChemToolDataset, the first large-scale chemical tool trajectory dataset. Simultaneously, we propose SMILES-GRPO to build a dense chemical reward function, promoting the model's ability to call chemical agents. Evaluations across diverse aspects of drug design show that ChemCRAFT outperforms current cloud-based LLMs in molecular structure analysis, molecular optimization, and synthesis pathway prediction, demonstrating that scientific reasoning is not solely an emergent ability of model scale, but a learnable policy of tool orchestration. Furthermore, assessments in real-world drug optimization scenarios, including fentanyl analog optimization and pro-inflammatory GSTO1 inhibitors optimization, reveal that ChemCRAFT generates optimization trajectories mirroring human experts, confirming its reliability as a scientific assistant. This work establishes a cost-effective and privacy-preserving paradigm for AI-aided chemistry, opening new avenues for accelerating molecular discovery with locally deployable agents.

Research Square
National University of Singapore (SG), Peking University (CN)
Peking University
Computational Drug Discovery Methods
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon