Loading…
A Beijing research institute working on core AI methods and their application to the natural and social sciences, from theory to industry deployment.
The Zhongguancun Institute of Artificial Intelligence (ZGCI, 中关村人工智能研究院) is a research institute in Beijing's Haidian district that organizes its work along three axes: AI core research (model architectures, optimization, reinforcement learning, AI safety, embodied intelligence, and infrastructure), AI for the natural sciences (physical sciences, life sciences, medicine, and scientific computing), and AI for the social sciences (education, finance, governance, and ethics). It operates in close partnership with Beijing Zhongguancun Academy, sharing facilities and research staff, and positions itself as a bridge between fundamental method development and deployed applications.
Tongji University / Hong Kong University of Science and Technology / East China Normal University / Anqing Normal University / New York University Shanghai / Monash University / Zhongguancun Institute of Artificial Intelligence
Released September 6, 2026
Protein-sequence-conditioned mixture-of-experts language model that designs small molecules and emits an explicit synthesis route for each one.
Beijing Zhongguancun Academy / Zhongguancun Institute of Artificial Intelligence / Tsinghua University / Peking University / Georgia Institute of Technology
Released September 23, 2025
Hierarchical transformer for virus discovery in metagenomes, classifying viral genomes across taxonomic ranks and flagging candidate novel lineages.
Alibaba Cloud / Beijing Zhongguancun Academy / Zhongguancun Institute of Artificial Intelligence / University of Science and Technology of China / Agricultural Genomics Institute at Shenzhen / Hong Kong University of Science and Technology (Guangzhou) / Hong Kong University of Science and Technology / Mila / Université de Montréal / HEC Montréal / Carnegie Mellon University
Released February 11, 2025
Long-context generative genomic foundation model with a 98k-nucleotide window, trained on 386 billion bases of eukaryotic DNA for sequence design.