China unveils AI model to revolutionize soybean breeding

August 17, 2026 (MLN): Chinese researchers have launched Fengshu 2.0, an upgraded artificial intelligence model designed specifically for soybean research and breeding, at a national soybean research symposium, according to Anhui Agricultural University in Hefei, Anhui Province.

Developed by Anhui Agricultural University and the Institute of Crop Sciences under the Chinese Academy of Agricultural Sciences, the model combines extensive soybean-related datasets, including germplasm resources, genomic and coding sequences, proteins, transcripts, gene expression information, disease samples, breeding trial records and domestic and international research literature.

Fengshu 2.0 features a multimodel collaborative system that connects several general-purpose large language models to handle specialized soybean research queries.

A dedicated module compares and consolidates their responses using domain knowledge, proprietary datasets and structured evaluation rules, assessing areas including consensus, differences and evidence completeness, according to APP.

The upgraded platform can support soybean disease diagnosis, breeding-parent selection, virtual cross design, phenotypic data analysis, molecular marker-assisted breeding and candidate gene function analysis.

The original Fengshu model was introduced in February as a generative AI breeding platform covering the soybean life cycle. The latest upgrade is aimed at shifting breeding from traditional experience- and experiment-driven approaches toward data-supported decision-making.

Researchers said Fengshu 2.0 can improve data processing efficiency, reduce repetitive tasks and help optimize breeding strategies.

The team plans to integrate climate, soil and cropping-pattern data from major soybean-producing regions in China to support seed development, cultivation, research and agricultural extension.

Future development will focus on expanding field trials, strengthening professional datasets and refining algorithms while continuously validating AI-generated results against real-world breeding experiments.

The platform is also expected to improve capabilities in germplasm discovery, trait assessment, variety selection and production management, supporting the development of soybean varieties with higher yields, improved quality and greater stress tolerance.

Copyright Mettis Link News