China open-source AI usage hits record high as DeepSeek leads
China's use of open-weight AI models has reached a new high, surpassing proprietary models for the first time. Data from US web development platform Vercel indicates that on August 25, Chinese open-weight AI models accounted for 54% of total token usage across its calls and the AI Gateway agent layer. This shift reflects a growing preference for open-source solutions within China's AI development landscape, with DeepSeek emerging as a leading player in this segment. The trend suggests a maturing domestic open-source AI ecosystem. This development could reshape how AI innovation and deployment proceed in the region.
The surge in open-weight AI model usage in China, now exceeding proprietary models at 54% of token usage on Vercel as of August 25, marks a significant inflection point. This data, cited by SCMP, points to a robust domestic open-source AI ecosystem, with DeepSeek at the forefront. The development suggests Chinese developers are increasingly adopting and contributing to open-source frameworks, potentially accelerating localized AI innovation and reducing reliance on foreign proprietary technologies. For Asia, this shift could foster greater collaboration and competition within the regional AI sector. It also presents an opportunity for Chinese companies to build on open foundations, customizing models for specific market needs without the licensing constraints of proprietary alternatives. The thing to watch is how this trend impacts the commercialization strategies of both open-source contributors and proprietary model developers in China and across Asia.
分享這篇文章
相關文章
6 則
Alibaba pushes into South America’s AI market with launch of Brazil data centres
Alibaba's expansion into South America's AI market shows the global reach of Asian AI players.

美國抑或中國?哈薩克發展人工智能拒絕選邊站

LUXEED 與 Aimoga 機器人:透過互聯的人車家生態系統共同建構未來出行

馬來西亞希望擺脫Waze。司機們會轉用嗎?

Stripe 發票不等於收入:如何透過網絡掛鈎將其轉化為 ASC 606 附表

