Reference News Network, July 31 report - According to the Russian FinaM Information Agency website on July 29, the cost of using China's latest artificial intelligence (AI) model is lower than DeepSeek from DeepSeek Company.

According to U.S. media reports, the startup company Beijing Zhipu Hua Zhang Technology Co., Ltd. released its new model GLM-4.5 on the 28th, which has a lower calling price than DeepSeek.

GLM-4.5 is a foundational model designed for agent applications. Zhipu stated that the new model will be open-sourced simultaneously on multiple platforms, allowing developers to download and use it for free.

Additionally, according to the U.S. Consumer News and Business Channel website on July 28, Chinese companies are developing more intelligent and less expensive artificial intelligence models, showing similarities with DeepSeek's market-shaking breakthroughs in many key aspects.

On the 28th, the startup company Beijing Zhipu Hua Zhang Technology Co., Ltd. released its new artificial intelligence model GLM-4.5, and published related data showing that its API (Application Programming Interface) calling price is lower than DeepSeek. The new model also adopts an open-source model, meaning developers can download and use it for free.

A company executive said in an interview that the parameter count of GLM-4.5 is approximately half that of DeepSeek-R1.

The report stated that earlier this year, DeepSeek caused a shock in the global investment community. Its developed artificial intelligence model not only has performance comparable to OpenAI's ChatGPT, but even the training and operating costs are lower than ChatGPT.

Data released by Zhipu shows that the API calling price of GLM-4.5 is as low as 0.8 yuan per million tokens for input and 2 yuan per million tokens for output. Token is a unit of measurement for data processing in artificial intelligence models.

According to the report, the company, established in 2019, is preparing for an initial public offering.

Earlier this month, the AI startup Moonshot AI, supported by Alibaba, released the Kimi K2 model, which is claimed to outperform OpenAI's ChatGPT and Anthropic's Claude in certain capabilities.

In recent weeks, several Chinese tech companies have released open-source large models. During the 2025 World Artificial Intelligence Conference held in Shanghai, Tencent launched the Hunyuan 3D World Model 1.0; recently, Alibaba also released the programming large model Qwen3-Coder.

The Japan Economic News website reported on July 26 that the number of AI models registered domestically in China has significantly increased over the past six months, showing a desire to become a global leader.

In terms of the performance of artificial intelligence models, the gap between China and the United States is narrowing. A report released by Stanford University earlier this year showed that in tests where the accuracy of dialogue-based AI answers was determined by human voting, top AI products in the United States had a 9.26% score advantage over Chinese products in January 2024, while the gap had narrowed to 1.7% by February 2025.

Earlier this month, Huang Renxun, CEO of semiconductor giant NVIDIA, visited China, pointing out that China has cultivated about 50% of the world's AI technology researchers, with a strong talent pool in science, mathematics, and computer fields, forming an active ecosystem.

An increasing number of Chinese companies are turning their attention to open-source AI. At the beginning of the year, DeepSeek, an open-source model from DeepSeek, made a stunning debut and attracted worldwide attention. Companies such as Alibaba are also accelerating the release of open-source large models. Open-source models allow more developers to participate in technical improvements, thereby accelerating performance improvements.

KPMG Financial Consulting's executive Takahashi Jun pointed out that although the United States is striving to maintain an advantage in the technological field, under national leadership, the connections between China's AI-related industries are constantly strengthening, and the rising momentum in some areas and uses is significant. (Translated by Hu Guanghe, Han Ziling, Ma Xiaoyun)

Original: https://www.toutiao.com/article/7533057550896349731/

Statement: This article represents the views of the author. Please express your opinion by clicking on the [up/down] buttons below.