Thursday, June 25, 2026

North Korea Opens Door to Russian Tourists, Eyes Chinese Visitors Next

North Korea-China exchanges are reviving, with hopes for renewed tourism as rail links and coastal zones open to visitors.

China Outpaces U.S. in Quantum Technology Investment, Boosting Its Global Ambitions

China invests heavily in quantum technology, challenging the U.S., while South Korea aims to enhance its quantum capabilities.

North Korea’s 3rd Training Session for City and County Party Secretaries: Key Insights and Outcomes

North Korea's 3rd training session for city and county party secretaries focused on local development and implementing the Party's agenda.

Z.ai’s GLM-5.2 Surpasses Google Gemini: What This Means for AI in Asia

TechZ.ai's GLM-5.2 Surpasses Google Gemini: What This Means for AI in Asia
/ Image created by AI tool related to article content
/ Image created by AI tool related to article content

Chinese artificial intelligence (AI) startup ZhipuAI (Z.ai) is making waves in the AI industry with its open-source model GLM-5.2, which has surpassed Google’s Gemini in performance.

Industry reports from Thursday reveal that GLM-5.2 scored an impressive 51 points in the latest AI performance evaluation by Artificial Analysis (AA), securing the fourth position. This places it just behind Anthropic’s Claude Fable 5 (60 points), Claude Opus 4.8 (56 points), and OpenAI’s GPT-5.5 (55 points).

While GLM-5.2 maintains the same model size as its predecessor GLM-5.1, with 744 billion total parameters and 40 billion active parameters, it has achieved significant improvements. The model has reduced hallucinations and enhanced scientific reasoning capabilities, contributing to its high score.

The industry was taken aback by these results, as GLM-5.2 not only outperformed Google’s latest AI model, Gemini 3.5 Flash (50 points), but also became the first Chinese AI to break the 50-point barrier. Moreover, it set a new record as the first open-source model to reach the 50-point range.

GLM-5.2’s cost-effectiveness is also turning heads. Its application programming interface (API) call price is remarkably competitive, with output token costs at just 4.40 USD per million tokens. This is a fraction of the cost compared to its rivals: Fable (50 USD), Opus 4.8 (25 USD), and GPT-5.5 (30 USD), ranging from 8.8% to 17.6% of their prices.

In the lucrative field of AI coding, GLM-5.2 demonstrated superior performance. It scored 62.1 points in the SWE-Bench Pro coding test, outperforming GPT-5.5’s 58.6 points.

ZhipuAI’s release of this high-performance open-source AI model has coincided with a surge in its corporate value. This comes amid recent actions by the U.S. administration to restrict overseas access to advanced AI models like Anthropic’s Fable 5 and Mythos 5.

As of June 24, ZhipuAI’s stock price reached 2,174 HKD (about 277 USD), marking a staggering 1,553% increase since its January listing.

ZhipuAI, established in June 2019 at Tsinghua University’s Computer Science and Technology Department, was founded by professors Tang Jie and Li Juanzi. The company has attracted investments from Chinese tech giants like Alibaba, Tencent, and Xiaomi, and recently secured a substantial 200 billion KRW (about 129 million USD) investment from a Shanghai state fund.

In response to Elon Musk’s prediction on X that China would develop an AI model comparable to Fable 5 by the first quarter of 2027, ZhipuAI founder Tang Jie confidently stated that it won’t take that long.

Check Out Our Content

Check Out Other Tags:

Most Popular Articles