The Chinese company Z.ai, formerly known as Zhipu AI, develops artificial intelligence models that approach the performance of leading American counterparts and in some aspects even surpass those of Anthropic and OpenAI.
The Chinese company focuses on developing a series of open GLM models, whose key feature is their ability to be deployed within the infrastructure of organizations that possess the necessary computing resources, and to be used independently of external services. This approach simplifies integration processes, but it also raises security challenges.
In contrast, OpenAI models are only available via an application programming interface (API), which imposes a number of limitations, particularly on users in Russia.
The latest model in the series, the GLM-5.2, is approaching the capabilities level of Claude Opus 4.8 and GPT-5.5. However, the model remains open and available through a number of service providers, including providers in the Russian market, although it still lags slightly behind the latest OpenAI and Anthropic models in some tasks.
In contrast, Chinese models achieve strong results in some areas, with GLM-5.2 outperforming its Western competitors in specific scenarios. However, Valentin Malich, head of the Big Language Model Development Center at MWS AI, pointed out that benchmarks only reflect limited aspects of model performance and do not provide a comprehensive picture of their quality.
Malikh said: "Based on practical experience, both from my own experience and that of my colleagues, it can be said that the Chinese model is still slightly behind the advanced models of OpenAI and Anthropic. As for whether the next generation of GLM will take the lead, it is not possible to say for sure, but the pace of development in this sector makes this scenario plausible."
He added that catching up with the leaders in the artificial intelligence market remains difficult, both financially and regulatoryly, given their vast capital and accumulated investments. He believes that changing the competitive landscape requires a radical technological breakthrough capable of reshaping the market.
Dmitry Borozdin, co-founder and CEO of RetailCRM, said that overtaking the leading companies may require moving to entirely new technological architectures, such as biomaterial-based solutions for energy-efficient neural connections or quantum processors, stressing that massive investments alone are not enough to overtake the companies that currently lead the market.
Andrey Sharipov, product manager at Fedot.ai, a subsidiary of digital logistics operator versta.io, noted that Z.ai's superiority over American models will remain a challenge in terms of capabilities, but he believes the competition may be decided from the value-for-price perspective, explaining that most users will not be willing to pay many times the cost for a slight improvement in performance, which gives Chinese models a competitive advantage in terms of economic efficiency.
