Core Insights - The core focus of the article is the launch of the OneScience platform by Zhongke Shuguang at the 2025 China Computer Conference, which aims to enhance the efficiency of scientific model development through an integrated environment for artificial intelligence-driven research [2]. Group 1: Platform Features - OneScience is designed based on an open architecture for artificial intelligence computing, leveraging domestic computing power to replicate and integrate numerous AI-driven scientific research models and datasets [2]. - The platform supports various fields such as Earth sciences, bioinformatics, fluid simulation, and materials chemistry, enabling users to develop new models and significantly improve research and optimization efficiency [2]. - It provides a one-stop development environment for scientific models, integrating scattered models and research scenarios into a unified intelligent research platform, covering the entire process from model training to inference application [2]. Group 2: Market Impact - The transition of AI scientific models from "task-oriented" to "general-purpose" is leading to rapid growth in model parameters, with the performance of the "super-intelligent fusion" system doubling every nine months [2]. - OneScience platform is currently available for download through supercomputing internet and various community platforms, supporting multiple domestic research institutions in efficient development across fields like biomolecular design and meteorology [2].

Core Viewpoint - The launch of OneScience, a one-stop development platform for scientific large models by Zhongke Shuguang, marks a significant advancement in China's computing industry, particularly in the realm of domestic GPU support [1] Group 1 - On October 23, Zhongke Shuguang unveiled its first scientific large model development platform, OneScience, at the 2025 China Computer Conference [1] - The OneScience platform supports model code reproduction and optimization on domestic GPUs, enhancing the capabilities of local developers [1] - The platform is accessible for download through various community platforms, including Github, Gitee, and supercomputing internet, promoting wider industry adoption [1]