Is This The New ChatGPT Killer? Cheap Chinese AI Model Stuns Silicon Valley

A Chinese artificial intelligence model is drawing attention across the technology industry after its performance surprised developers and executives in the United States.

Since DeepSeek shook global markets with a low-cost AI model last year, many businesses have viewed the market as a choice between cheaper Chinese systems with fewer capabilities and premium offerings from OpenAI and Anthropic, which have invested billions in development.

That perception is now being questioned following the launch of GLM-5.2 by Beijing-based startup Z.ai. The model has gained recognition for its coding abilities and agent capabilities, allowing it to complete complex tasks with minimal prompting while costing far less than many leading Western alternatives. Its growing popularity has prompted fresh debate over whether Chinese AI companies are beginning to close the gap with their American rivals, particularly as questions continue over regulation, development costs and the future direction of the AI industry.

Rising Interest In GLM-5.2

GLM-5.2 was launched by Beijing-based startup Z.ai last month and has quickly become one of the most talked-about AI models among developers. The model has attracted attention for its coding performance and agent capabilities, which allow it to carry out complicated tasks with limited instructions.

Its popularity has grown rapidly on OpenRouter, a third-party platform used by AI developers, where it has climbed above Anthropic's models in the usage rankings. The model has also earned praise from several well-known figures in the technology industry, including Snowflake chief executive Sridhar Ramaswamy and venture capitalist Marc Andreessen.

David Sacks, who previously served as U.S. President Donald Trump's AI czar, also praised the model during the All-In podcast. He said, 'We now have a Chinese open-weight model that is as good as the currently available models from OpenAI and Anthropic.'

Sacks added that GLM-5.2 was 'just a tick below Opus 4.8 (from Anthropic) and right up there with GPT 5.5 (from OpenAI),' before warning that 'we cannot afford to do things that slow our companies down.'

Some industry observers believe demand for GLM-5.2 has also benefited from Washington lifting restrictions on Anthropic's Fable and Mythos models and the delayed public rollout of OpenAI's GPT-5.6.

Brian Tse, founder and chief executive of Beijing-based AI consultancy Concordia AI, said the international developer community is becoming more cautious about depending entirely on proprietary American AI models.

'The international developer community is increasingly aware that relying solely on proprietary, U.S.-based API models carries significant risk,' he said.

GLM-5.2 currently sits fifth on Artificial Analysis' large language model intelligence leaderboard and ranks second on Code Arena's front-end coding rankings. According to the rankings, it operates at roughly one-sixth of the cost of closed American frontier models such as Claude and the GPT series.

Z.ai has not disclosed how much it spent developing GLM-5.2. However, founder Tang Jie wrote on X last month that the company could produce a model on par with Anthropic's Fable before the first quarter of next year.

Adoption Challenges Remain

Despite the enthusiasm surrounding GLM-5.2, experts believe convincing large American businesses to adopt the model will remain a difficult task because of concerns about data security.

Tiezhen Wang, the former Asia-Pacific lead at Hugging Face, said GLM-5.2 makes open-source AI much easier to use because developers no longer need extensive fine-tuning before deployment.

'You just deploy the model, and without doing any complex fine-tuning, it is in a highly usable, ready-to-use state,' Wang said. 'This drastically lowers the barrier to entry for open-source adoption.'

However, Wang also noted that upgrading enterprise AI systems usually takes several months, making rapid adoption less likely.

Wei Sun, principal AI analyst at Counterpoint Research, said discussions have already begun among some European companies about whether GLM-5.2 could be used in business environments. Even so, she believes resistance remains in regulated sectors.

'In the EU and U.S., some clients, partners and regulated industries may simply be unwilling to accept Chinese models in their AI stack, regardless of technical performance or price,' she said.

A report from the non-profit RAND found that Chinese large language models increased their global market share from 3 per cent to 13 per cent during the two months after DeepSeek released its R1 model in January last year. The biggest increases came in developing countries and nations with closer political and economic ties to Beijing.

Not everyone believes security concerns will prevent wider use. Some experts argue that businesses can protect their information by running Chinese models through American cloud providers or on their own servers.

Poe Zhao, founder of the Hello China Tech newsletter, said developers are generally more interested in practical performance than where a model originates.

'Developers tend to care less about where a model comes from than whether it works, how much it costs and whether they can deploy or access it reliably,' Zhao said.

He added that the likely outcome would not be an immediate replacement of OpenAI or Anthropic, saying, 'The likely pattern is partial routing, not overnight replacement of OpenAI or Anthropic. So yes, it is a mini DeepSeek moment but in a narrower, developer-centric sense.'