Ever since DeepSeek rattled financial markets early last year with a powerful yet affordable AI model, the technology world has operated under a familiar assumption: Chinese AI tools were cheaper but less capable, while American giants like OpenAI and Anthropic remained the gold standard — backed by billions in investment.
That assumption may now be shifting. A new model called GLM-5.2, released last month by Beijing-based startup Z.ai, is generating significant buzz in Silicon Valley for its coding abilities and what developers call “agentic” capabilities — meaning it can carry out complex, multi-step tasks with very little instruction from the user.
Some in the tech world are already calling it a “mini DeepSeek moment.”
On third-party AI developer platforms like OpenRouter, GLM-5.2 has rapidly climbed the usage charts, now ranking above models from Anthropic. High-profile figures including Snowflake CEO Sridhar Ramaswamy and venture capitalist Marc Andreessen have praised its performance.
David Sacks, who previously served as U.S. President Donald Trump’s AI czar, spoke about the model on the All-In podcast last week. “We now have a Chinese open-weight model that is as good as the currently available models from OpenAI and Anthropic,” he said. He described GLM-5.2 as “just a tick below” Anthropic’s Opus 4.8 and “right up there” with OpenAI’s GPT 5.5, warning that “we cannot afford to do things that slow our companies down.”
His comments came just before Washington lifted restrictions on Anthropic’s Fable and Mythos models on Tuesday. Experts say those curbs, along with delays in rolling out OpenAI’s latest GPT-5.6 model to the public, have helped drive global interest in the Chinese alternative.
“The international developer community is increasingly aware that relying solely on proprietary, U.S.-based API models carries significant risk,” said Brian Tse, founder and CEO of Concordia AI, a Beijing-based consultancy focused on AI safety.
According to performance rankings, GLM-5.2 currently sits in fifth place on Artificial Analysis’ large language model intelligence leaderboard, which evaluates AI systems across benchmarks including reasoning and coding. It holds second place on Code Arena’s front-end coding rankings — a measure of how well AI generates websites and web applications — all while operating at roughly one-sixth the cost of leading closed-source American models like Claude and the GPT series.
Z.ai has not revealed how much it spent building GLM-5.2. The company, also known as Zhipu AI, declined to comment for this story. Anthropic and OpenAI did not respond to requests for comment.
In a message posted to X last month in response to Elon Musk, Z.ai founder Tang Jie said the startup expects to produce a model on par with Anthropic’s Fable before the end of the first quarter of next year.
Tiezhen Wang, a former Asia-Pacific lead at Hugging Face — a platform widely used by developers working with open-source AI — said GLM-5.2 represents a meaningful shift. “The shift GLM-5.2 brings is that the open-source model has become a plug-and-play, out-of-the-box product,” Wang said. “You just deploy the model and without doing any complex fine-tuning systems, it is in a highly usable, ready-to-use state. This drastically lowers the barrier to entry for open-source adoption.”
Despite the excitement, significant obstacles remain before GLM-5.2 sees widespread adoption among American businesses. Data security concerns tied to the model’s Chinese origins have made many U.S. enterprises — especially in regulated sectors like banking and cybersecurity — reluctant to integrate it. Wang noted that migrating and upgrading enterprise AI systems typically takes several months.
Wei Sun, a principal AI analyst at Counterpoint Research, said European companies have begun discussing potential enterprise use. However, Sun cautioned: “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.”
A report released earlier this year by the non-profit RAND, drawing on website traffic data from 135 countries, found that Chinese large language models’ global market share jumped from 3% to 13% in the two months following the launch of DeepSeek’s R1 model in January last year — a release that triggered a broad technology selloff due to the stark contrast between DeepSeek’s low development costs and the massive AI infrastructure spending by Western companies. The usage gains were most pronounced in developing nations and countries with close political and economic ties to Beijing.
Some analysts argue that fears about Chinese AI model security are exaggerated, pointing out that running such models on U.S. cloud infrastructure or a company’s own servers can adequately protect data. While large corporations move cautiously, smaller businesses and tech startups appear to be adopting the technology more quickly.
“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,” said Poe Zhao, a China tech analyst and founder of the Hello China Tech newsletter.
Zhao added that the more likely outcome is gradual adoption rather than a sudden replacement of OpenAI or Anthropic. “The likely pattern is partial routing, not overnight replacement of OpenAI or Anthropic,” he said. “So yes, it is a mini DeepSeek moment but in a narrower, developer-centric sense.”
