Beijing, January 22, 2026 - In a landmark move for domain-specific AI, researchers at Nanjing Agricultural University have released Sinong, China’s first completely open-source large language model tailored exclusively for agriculture. Named after the ancient Chinese god of farming, Sinong represents a deliberate push to bring advanced AI capabilities to one of the world’s most critical sectors.

The model was pretrained on a massive agriculture-focused corpus: more than 9,000 specialized books, 240,000 peer-reviewed papers, thousands of government policy documents, and extensive practical farming guides covering crops, livestock, soil science, and climate adaptation. This curated dataset gives Sinong deep domain knowledge that general-purpose models often lack when applied to nuanced agricultural queries.

Early benchmarks shared by the team show Sinong achieving over 98 percent accuracy on specialized tasks such as pest diagnosis from symptom descriptions, fertilizer recommendation based on soil composition, and interpretation of complex agricultural subsidies. Unlike proprietary vertical models, Sinong is fully open — weights, code, and training details are available on ModelScope and GitHub under permissive licenses.

The release aligns with China’s broader national strategy to modernize agriculture through AI, especially as climate change and labor shortages strain traditional farming. By open-sourcing the model, the developers hope to spur global collaboration, particularly in developing countries where smallholder farmers could benefit most from localized fine-tuning.

Dr. Li Wei, lead researcher on the project, stated: “General LLMs can answer farming questions, but they often hallucinate dangerous advice. Sinong is grounded in verified agricultural science and real-world policy, making it trustworthy for practitioners.”

Industry observers note that Sinong could accelerate the trend toward vertical LLMs, following recent specialized models in medicine and law. Its immediate availability lowers barriers for startups and cooperatives to build applications like chat-based crop advisors or automated policy compliance tools.

Credit: Nanjing Agricultural University research team led by Dr. Li Wei. Primary sources: EastFruit, AI + Tech Hub, ModelScope and GitHub repositories.