Beijing: Months after DeepSeek's dazzling debut, China's generative AI firms continue racing with unabated fervor, demonstrating that competition in this sector remains dynamic, with vast innovation potential yet to be realized.
According to Namibia Press Agency, this month saw a parade of Chinese AI foundation models unveiling major updates. Moving beyond the "chatbot" label, these open-source models are evolving into smart agents designed to solve complex, real-world problems. Their sustained emergence, rivaling or even surpassing global top competitors, underscores an enduring creative vitality in China's AI landscape.
On Monday, Beijing-based Zhipu.AI launched GLM-4.5, a versatile and efficient AI platform integrating reasoning, coding, and agent capabilities. "The model now creates sophisticated standalone artifacts, ranging from interactive mini-games to physics simulations," stated the update on Zhipu's official X account. The launch was met with positive reactions, including from Daniel Vila Suero of Hugging Face, a platform for the global AI community. U.S. tech media website VentureBeat praised GLM-4.5 as a "truly viable, high-performance foundation model" for enterprise teams.
Alibaba's AI division also recently updated its Qwen3 reasoning model, with the team promising even more significant updates in the future. Beijing-based startup Moonshot AI released Kimi K2, an open-source model praised by Nathan Lambert of the Allen Institute for AI as "the new best-available open model by a clear margin."
Shanghai's StepFun and MiniMax have introduced Step-3 and M1 models, respectively, with Step-3 noted for its ability to be trained on home-grown semiconductors and M1 employing low-cost training methods. Shenzhen-based Tencent debuted HunyuanWorld-1, capable of creating fully immersive 360-degree virtual worlds.
The open-source model leaderboard on Hugging Face is currently led by GLM-4.5, with Qwen and Tencent models close behind. K2 ranks ninth, highlighting the strong presence of Chinese models among the top performers. Industry experts, such as Nvidia's CEO Jensen Huang, recognize the global impact of these developments, noting that Chinese models are world-class and have spurred AI advancements worldwide.
China has launched 1,509 AI models out of the 3,755 globally released, leading all countries in AI model launches. The nation also hosts 71 of the world's 271 AI unicorns, according to the China Academy of Information and Communications Technology. The Chinese government recently proposed creating a global AI cooperation organization to foster worldwide tech collaboration.
Du Yulun, a researcher at Moonshot AI, expressed optimism for China's AI contributions to the open-source community, stating, "We look forward to China's AI giving back even more to the open-source community -- so innovators everywhere can advance together."