China’s artificial intelligence ambitions have received a fresh boost after Zai founder Tang Jie suggested that the country could develop a Mythos-class AI model much sooner than many industry observers expect.
His comments followed the launch of Zai’s newest model, GLM 5.2, which has already proven to be one of the highest-performing open-weight AI systems globally.
The remarks by Tang have gathered interest owing to the growing technological competition between the US and China for advanced AI systems.
According to posts shared on X, a user speculated that China could develop a fully domestic Mythos-level model by the end of 2026. Tesla and xAI chief Elon Musk responded by suggesting such a milestone was more likely by the first quarter of 2027. Tang, however, disagreed, replying that it “won’t take that long.”
Also Read | GTA 6 pre-orders open on June 25 as Rockstar confirms November 19 release date
US restrictions add new dimension to AI competition
The argument emerges in light of the latest restrictions implemented by the US on foreign use of Anthropic’s Mythos 5 and Fable 5 models. As per media reports, the action came after fears arose that China-affiliated parties might take advantage of sophisticated AI technologies for cyber activities.
The Mythos class models are renowned for their capability to detect weaknesses in software and execute complex cybersecurity procedures. Analysts consider such systems as some of the most sophisticated AI technologies available currently.
Should China-based companies be successful in coming up with similar technologies, it is expected that the effect of restrictions imposed by the US will be lessened.
GLM 5.2 strengthens China’s AI ambitions
Zai’s latest model, GLM 5.2, has become a major talking point within the AI community following strong benchmark results. According to Artificial Analysis rankings, GLM 5.2 currently leads the open-weight AI category with a score of 51, outperforming several prominent models, including Google’s Gemini 3.5 Flash and Anthropic’s Claude Sonnet 4.6.
The model remains behind Anthropic’s Fable 5 and OpenAI’s GPT 5.5 but has narrowed the performance gap significantly.
One of the most notable aspects of GLM 5.2 is its development process. The model was reportedly trained entirely using Huawei Ascend chips without relying on Nvidia hardware, a milestone viewed as significant given US export restrictions on advanced semiconductor technology
Industry estimates cited by AI entrepreneur Emad Mostaque suggest training costs for GLM 5.2 may have been substantially lower than those associated with leading American AI models.
Questions over model development remain
The rapid progress of Chinese AI companies has also reignited debate around model distillation, a technique where smaller models are trained using outputs generated by larger systems.
Also Read | Fed up with algorithms? Meta offers more control over your social media feed
Several US-based AI companies have previously alleged that some Chinese firms used distillation methods to accelerate development, though such claims continue to be debated within the industry.
As global competition intensifies, experts believe the race to build increasingly capable AI systems is likely to remain a key battleground in the broader technology rivalry between Washington and Beijing.
FAQs
1. What did Zai founder Tang Jie say about China’s AI development?
Tang Jie suggested that China could develop a Mythos-class AI model before the first quarter of 2027, indicating progress may happen sooner than expected.
2. Why is GLM 5.2 significant?
GLM 5.2 is one of the highest-performing open-weight AI models globally and was reportedly trained entirely on Huawei Ascend chips, highlighting China’s growing AI capabilities despite US technology restrictions.
