ChemGraph as an agentic framework for computational chemistry workflows
ChemGraph как агентная платформа для рабочих процессов вычислительной химии
2026-01-08
SCID: 54.1/mydnbjsh
Discuss with AI
ChemGraphagentic frameworkcomputational chemistry workflowsgraph neural network-based foundation modelsmulti-agent framework
Figures from the paper
Abstract (AI)
Atomistic simulations are essential in chemistry and materials science but remain challenging to run due to the expert knowledge required for the setup, execution, and validation stages of these calculations. We present ChemGraph, an agentic framework powered by artificial intelligence and state-of-the-art simulation tools to streamline and automate computational chemistry and materials science workflows. ChemGraph leverages graph neural network-based foundation models for accurate yet computationally efficient calculations and large language models (LLMs) for natural language understanding, task planning, and scientific reasoning to provide an intuitive and interactive interface. We evaluate ChemGraph across 13 benchmark tasks and demonstrate that smaller LLMs (GPT-4o-mini, Claude-3.5-haiku, Qwen-2.5-14B) perform well on simple workflows, while more complex tasks benefit from using larger models. Importantly, we show that decomposing complex tasks into smaller subtasks through a multi-agent framework enables GPT-4o to reach perfect accuracy and smaller LLMs to match or exceed single-agent GPT-4o's performance in these benchmarks.
Key Findings
1
Across 13 benchmark tasks, smaller LLMs perform well on simple workflows, whereas larger models are advantageous for more complex tasks.
2
ChemGraph is an agentic framework that automates computational chemistry and materials science workflows using AI and state-of-the-art simulation tools.
3
Decomposing complex workflows into subtasks with a multi-agent framework enables GPT-4o to achieve perfect benchmark accuracy.
4
Graph neural network-based foundation models provide computationally efficient calculations, while large language models handle natural-language interaction, planning, and scientific reasoning.
5
The multi-agent approach allows smaller LLMs to match or exceed the performance of single-agent GPT-4o on the evaluated benchmarks.
Research Object
ChemGraph agentic framework for computational chemistry and materials science workflows
Research Subject
automation, task-planning, scientific reasoning, and benchmark performance of AI-assisted atomistic simulation workflows
Publication Details
Publication Date
2026-01-08
Journal
Publisher
ISSN
Cited by
17
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai7
Commentary: The Materials Project: A materials genome approach to accelerating materials innovation2013
The ORCA quantum chemistry program package2020
The atomic simulation environment—a Python library for working with atoms2017
Extended tight‐binding quantum chemistry methods2020
Computational approaches streamlining drug discovery2023
Augmenting large language models with chemistry tools2024
Applying Classical, Ab Initio , and Machine-Learning Molecular Dynamics Simulations to the Liquid Electrolyte for Rechargeable Batteries2022