China’s Z.ai Challenges OpenAI, Anthropic With New GLM-5.3 Coding Model

Chinese AI firm Z.ai Co. has launched a flagship AI model, GLM-5.3 with coding capabilities. The move adds to the list of Chinese open-weight startups that are taking on frontier models developed by Anthropic and OpenAI.

Z.ai, which is also known as Zhipu, plans to launch weights of the upgraded GLM-5.3 AI coding model in the coming two weeks. Weights include the underlying parameters that enable developers to download and customize an AI system. The Chinese AI firm released GLM-5.3 days after NVIDIA unveiled Nemotron 3.5 Lightning, an open-weight AI model for multi-agent systems.

How Does GLM-5.3 Perform in AI Tests?

According to the benchmark scores shared by the company, the new Z.ai AI model performs better than GLM-5.2. The scores showed the new model’s performance closes in on, and sometimes surpasses that of Anthropic’s Fable 5. Z.ai said GLM-5.3 is built on the same base model as GLM-5.2, which the company released in June. However, the approximately 700-billion-parameter model received expanded training to improve its performance on complex, long-running coding and agentic tasks. Chinese AI models are increasingly ranking highly in artificial intelligence tests. DeepSeek’s latest V4-Flash AI model scored 50/100 on the Intelligence Index, which is comparable to Gemini 3.6 Flash. Moonshot AI’s Kimi K3 model performed better than OpenAI’s GPT-5.5 and Anthropic’s Claude Opus 4.8 in benchmarks like general agents and coding. However, it trails GPT-5.6 Sol and Claude Fable 5 in overall performance. With GLM-5.3’s coding capabilities, Z.ai is stepping up the competition. The AI firm plans to release the new model under a permissive license, which is designed to encourage wider adoption among users and developers.

Why Z.ai Matters in the AI Race

Z.ai has become a major player in China’s AI industry and was the first large language model developer globally to go public. After launching GLM-5.2, its valuation rose to about $137 billion, surpassing companies such as PDD Holdings and NetEase briefly. Its value has since dropped to around $80 billion. The AI firm reportedly completed construction of a data center in July. The facility is equipped with at least 10,000 Chinese-made AI chips to support the development and deployment of its GLM models. In the same month, the AI startup’s annual recurring revenue reached $1 billion, reflecting the growing demand for AI services from developers and businesses.

What GLM-5.3 Means for Z.ai

By upgrading GLM-5.3 with coding capabilities, Z.ai is positioning itself for a larger role in the global AI race as demand for capable, open-weight coding models grows. The move shows how quickly China’s AI sector is closing the gap with leading American model developers. If GLM-5.3 delivers on its benchmark gains and attracts developers, Z.ai could make the competition for AI leadership even more intense.