Shanghai is vying to create its own open-source artificial intelligence (AI) ecosystem, as the success of DeepSeek models have reshaped the landscape of the global AI competition.
Advertisement
At the Global Developer Conference, an AI community event hosted in Shanghai over the weekend, open-source developers from around China congregated in a show of exuberance over the possibilities of AI since DeepSeek’s resource-efficient models captured the world’s imagination. Use cases on display included everything from robotics to virtual reality glasses.
“We will further improve the open-source ecosystem, strengthen the service system for open data and open models, and continue building the open-source community,” Shanghai Vice-Mayor Chen Jie said on Saturday at the opening ceremony of the three-day event, which started on Friday.
The value of Shanghai’s AI industry surpassed 450 billion yuan (US$62 billion) in 2024, according to Chen, with 60 major models registered with the Cyberspace Administration of China (CAC). China requires that AI models be registered with the internet regulator before public release. The CAC had approved a total of 302 generative AI services by the end of last year.

Xiong Jijun, a vice-minister in the Ministry of Industry and Information Technology, also spoke at the event, emphasising the importance of open-source initiatives. “We support the development of open-source communities built on openness, co-creation, sharing and collaborative governance,” he said.
Advertisement
Open source was a central theme of the conference, which marked the first major industry gathering since the rapid rise of DeepSeek’s open-source large language models V3 and R1 in December and January. On Friday, the Hangzhou-based start-up said it would open five code repositories to the public.