(Yicai) Feb. 24 -- Experts from around the world gathered at the three-day 2025 Global Development Conference in Shanghai to explore new development opportunities brought into the artificial intelligence ecosystem by open-source large language models.

Breakthroughs in open-source LLM technologies and products have brought new opportunities and space for developing China's AI industry, Xiong Jijun, vice minister for industry and information technology, said at the opening ceremony of the GDC that ended yesterday.

China has transitioned from benefiting from open-source communities to becoming a contributor, noted Shen Xiangyang, chairman of the board of trustees at the Hong Kong University of Science and Technology.

Open source is crucial for developing AI as it enhances community cohesion, reduces costs, increases the utilization of AI technologies, and strengthens trust within the industry, according to Andrew Aitken, former global director of Linux Foundation's Fintech Open Source Foundation. AI investments are highly repetitive, so AI firms must balance profitability and commercial value, he added.

The open-source ecosystem attracts numerous partners, said

Jiang Daxin, founder of StepFun. On Feb. 18, the Shanghai-based firm open-sourced two video generation models and a voice model, and it plans to open-source an image-to-video product in March.

MiniMax, the owner of Hailuo AI, the world's largest video generation website, opened-sourced its next-generation 01 series of AI models in January. The company will continue to open-source its tech and promote the foundational ecosystem development of non-transformer architectures, noted Vice President Liu Hua.

In addition, the corpus dataset has attracted significant attention.

While big data open-source provides tools, the corpus is the key, just as an internal combustion engine needs diesel, said Huang Haiqing, chief executive of Kupasi Technology, the first national AI corpus company established last March. The firm has set up comprehensive and industry-specific corpora and plans to develop synthetic and chain-of-thought data, Huang added.

According to Pan Yan, director of the AI department of the Shanghai Municipal Commission of Economy and Informatization, the corpus is the key to competition, and the city has already made the necessary arrangements.

Shanghai will improve the open-source ecosystem and promote the sharing and openness of models, algorithms, and data resources, said Vice Mayor Chen Jie.

The global scientific intelligence developer community and the upgrade plan for the Mosu Space carrier were among some initiatives launched at this year's GDC.

Editor: Martin Kadiev