Chinese Z.ai Unveils GLM-5.3: An Open AI Model to Rival Anthropic and OpenAI Flagships

The Chinese company Z.ai, also known as Zhipu, has announced an upcoming flagship AI model, GLM-5.3, which will join its lineup of open products competing with American Anthropic and OpenAI. The new model, with approximately 700 billion parameters like its predecessor GLM-5.2 released in June, offers improved coding capabilities, aiming to close the gap with leading models, including Anthropic Fable 5. Z.ai plans to release the model's weights within two weeks, making it accessible to everyone. Published benchmark results indicate that GLM-5.3 will significantly outperform GLM-5.2 and may rival or exceed Fable 5 in certain tests. While OpenAI and Anthropic are considered global leaders in AI, Chinese companies like Moonshot Kimi, DeepSeek V4, and Alibaba Qwen have demonstrated comparable results at lower costs. Z.ai follows this trend by offering open-licensed products to attract users and developers. Meanwhile, DeepSeek has raised prices for its V4 Flash and Pro models, though expert estimates suggest its flagship V4 Pro remains cheaper than GLM-5.2. In independent evaluations, both DeepSeek's flagship and GLM-5.2 scored 53 points, trailing Kimi K3, Anthropic Claude Fable 5, and OpenAI GPT 5.6. Z.ai, one of China's AI pioneers and the first major AI developer to go public, saw its market capitalization reach $137 billion after the release of GLM-5.2, later settling at $80 billion. The company has completed a data center with over 10,000 Chinese AI accelerators for model training and deployment, and expects annual recurring revenue to reach $1 billion.