(Yicai) Oct. 13 -- The Chinese Academy of Engineering has recognized the open-source large language models developed by Chinese artificial intelligence startup DeepSeek and an AI chip designed by US tech firm Nvidia among the world’s top 10 engineering achievements.

DeepSeek’s two LLMs — DeepSeek-V3 and DeepSeek-R1 — launched between December last year and January this year, pioneered a new approach to exploring general AI under resource-constrained conditions through systematic innovations in model algorithms and engineering optimization, making DeepSeek one of the world’s fastest-growing AI developers, Engineering, a journal under the Chinese Academy of Engineering, announced today.

Nvidia’s Blackwell graphics processing unit architecture, introduced in March last year and designed to address the complex computational demands of AI inference, has significantly accelerated AI training, cut inference latency to seconds, reduced computational thresholds for ultra-large models, and helped usher AI systems into the trillion-parameter era, the journal noted.

Since 2021, Engineering has selected the world’s top 10 engineering achievements annually from major engineering and technological innovations completed within the past five years that have proven effective in practice and had a global impact. The list includes landmark projects, critical technical equipment, and original breakthroughs in engineering and technology.

This year’s list also features major tools for space and deep-sea exploration, including the US National Aeronautics and Space Administration’s Perseverance Rover, the European Space Agency’s Euclid Space Telescope, and China’s Full-Ocean-Depth Manned Submersible.

China’s South-to-North Water Transfer Middle Route Project, designed to optimize national water resource distribution, and the Taklimakan Desert “Edge-Locking” Project, an ecological barrier along the desert’s perimeter, were selected as major engineering projects. Other breakthroughs recognized include high-performance carbon fiber composite, humanoid robots advancing from lab to real-world applications, and antibody-drug conjugates for targeted cancer therapy.

Editors: Dou Shicong, Emmi Laine