BEIJING, April 28 (Xinhua) — The Chinese Academy of Sciences (CAS) on Tuesday unveiled an artificial intelligence (AI) model system to support scientific research and power multiple scientific fields, including mathematics, physics and biology. This marks a shift in AI-driven scientific research from fragmented, isolated exploration to collaborative, efficient, platform-based innovation.
The model system, named ScienceOne 100, is built on the science infrastructure model ScienceOne, which features clusters of multidisciplinary, domain-specific, large-scale models. ScienceOne provides three core features: Literature Compass, Innovation Assessment, and Agent Factory to power your entire research and innovation workflow.
ScienceOne was released in 2025 and was trained on specialized scientific corpora and data. Its latest version achieved flagship-level performance in terms of scientific knowledge and long-term reasoning by agents, and state-of-the-art results in terms of multiple benchmarks related to understanding and manipulating scientific images.
Literature compass provides in-depth reading of core literature and autonomous review writing, improving researchers’ work efficiency.
Innovation assessment can recognize cutting-edge dynamics in the scientific community and industry, allowing researchers to efficiently identify key scientific questions and potential innovation directions.
Meanwhile, Agent Factory enables autonomous closed-loop agent toolchains and intelligent assistance generation, provides over 2,000 research tools, and supports over 10 research domains.
The model system currently consists of eight domain-specific large-scale models covering mathematics, physics, materials science, astronomy, environmental science, aerospace, earth science, and biology.
The system has already been implemented and applied in more than 50 CAS institutions, covering more than 100 research scenarios. It has demonstrated tremendous potential in typical applications such as high-speed rail flow field reconstruction, spectral identification, material discovery, auxiliary design, astronomical observation, Qinghai-Tibet Plateau scientific expedition, ocean forecasting, and ecological research. ■
