The World Artificial Intelligence Conference started in Shanghai. Chinese company Z.ai introduced the GLM-4.5 AI model, cheaper than DeepSeek and built on “agentic” AI. The model is open sourced, needs only 8 Nvidia H20 chips, and charges less for input and output tokens compared to DeepSeek.

Z.ai’s new GLM-4.5 model challenges DeepSeek with lower costs. The company doesn’t need more Nvidia chips for now, but training costs are undisclosed. DeepSeek previously sparked controversy for defying U.S. chip restrictions. Other Chinese companies like Moonshot and Alibaba have also released competitive AI models.

OpenAI warned about Chinese AI progress, adding Z.ai to the entity list restricting U.S. companies. Z.ai, planning an IPO in Greater China, has raised over $1.5 billion from investors including Alibaba and Tencent. Other Chinese companies like Tencent and Alibaba have also released new open-source AI models for game development and coding.

Read more at CNBC: China’s latest AI model claims to be even cheaper to use than DeepSeek