Two weeks ago, the artificial intelligence company Anthropic was forced to shut down its two most powerful systems after an unexpected demand from the United States government to cut off access to that technology. The decision caught the market off guard and shook the industry in ways few had imagined, creating a vacuum nobody expected would be filled so quickly, let alone by a Chinese company. The entire sector was watching closely to see what would come next, and the answer did not take long to show up.
That is when Z.ai stepped onto the scene with impressive speed. Just days after the announcement about Anthropic’s Fable and Mythos models, the Chinese startup launched GLM-5.2, a model with performance remarkably close to the tools that had just been pulled from the American market. But with one twist that caught everyone in Silicon Valley off guard: an absurdly more competitive price tag. For certain specific tasks, the cost of using GLM-5.2 is up to 8 times lower than Anthropic’s Claude Opus 4.8, according to data from OpenRouter, and that did not go unnoticed by anyone.
The impact was almost instantaneous. The Z.ai model quickly climbed into the top 10 of the global ranking of most-used AIs, became a hot topic among developers, startups, and American investors, and sparked a wave of discussions about what is really happening in the race for artificial intelligence leadership worldwide. Worth noting that, of the models currently in that closely watched ranking, six were developed in China. Because this story goes well beyond a cheaper product showing up at the right time. It represents something much bigger. 🚀
What is Z.ai and why it matters right now
Z.ai is not exactly a newcomer in the artificial intelligence ecosystem, but for most of the Western public, it was still an unknown name until the launch of GLM-5.2. The company is known for developing the GLM series of models, short for General Language Model, and has been investing heavily in recent years to position itself as a real and robust alternative to the American giants in the sector. Headquartered in China, Z.ai operates within a regulatory and competitive landscape that looks completely different from what you see in the United States, which, interestingly enough, might be one of the reasons it can work with such different cost margins.
The GLM-5.2 represents a considerable technical leap compared to earlier versions of the series. Like most high-performance Chinese models, it is open-source software, meaning anyone can use and modify the technology for free. That makes usage much cheaper, even if the model is not quite as powerful as what American companies have built. As Vivek Ramaswami, investor at Madrona Venture Group, put it well, you do not always need to drive a Ferrari everywhere you go.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
The model stands out especially in computer code generation and in powering AI agents, those digital assistants capable of using other software to carry out tasks. According to Anastasios Angelopoulos, CEO of ArenaAI, which tracks millions of AI users, Z.ai’s technology is already the third most used in the world for artificial intelligence tasks. For anyone building applications that rely on AI at scale, that combination of performance and price is extremely hard to ignore. 🤖
Regulatory pressure and the vacuum that opened up in the market
When the American government demanded that Anthropic deactivate its two most advanced models, the artificial intelligence market felt the impact in a very direct way. Companies that depended on those tools for their products and services had to scramble for alternatives with urgency. This happened at exactly the moment when American companies realized they needed to find ways to cut how much they were spending on AI, and also when Silicon Valley executives started getting worried about the possibility of the government regulating the technology more strictly.
As Rehaan Ahmad, co-founder of the startup alphaXiv, pointed out, with Fable restricted, the gap between the United States and China got very small. He has been using Z.ai’s new model for over a week and noticed firsthand how the capability gap between the two powers keeps shrinking. This kind of regulatory interference is nothing new in the tech sector, but when it hits the artificial intelligence segment directly, the consequences spread much more broadly and rapidly.
Major cloud computing providers, including Microsoft and Amazon, already offer access to some systems from Z.ai, DeepSeek, MiniMax, and other Chinese startups. Microsoft itself even considered adding DeepSeek’s latest model as an option to power one of its products, which currently runs on technology from Anthropic and OpenAI. What makes this situation even more interesting is what it reveals about the fragility of the Western AI ecosystem in the face of regulatory decisions. The concentrated dependency on a handful of models and a handful of companies creates single points of failure that, when triggered, open up enormous space for new players to step in. 📊
The obstacles Chinese models still face
Despite all the enthusiasm, Chinese models still face two major challenges for broader adoption in the United States. The first is concern about these companies’ ties to the Chinese government. The second is complaints that Chinese companies may have used American technology unfairly to build these cheaper models. Even so, the low cost keeps winning over more and more supporters.
Z.ai was added to the U.S. Department of Commerce’s commercial blacklist in 2025. Corporate documents show that several of the company’s shareholders are controlled by a Chinese government agency that oversees the country’s defense industry. Because of that, some software developers are hesitant to use the system directly from computers in China, worried about sharing data with the company or with the government. There is also concern about Chinese efforts to censor their AI systems.
Still, companies can use the model without sending data back to China, as long as they are careful when setting up their systems, according to Wei Chen, general counsel at Infoblox. Chinese models do not have the same restrictions if you host them yourself or use another provider, Ahmad from alphaXiv explained, noting that right now there are actually more restrictions on Anthropic’s models than on the Chinese ones.
Accusations of copying and the distillation controversy
Anthropic and OpenAI have accused Chinese companies of improperly collecting data from their systems to speed up the development of Chinese technology. Anthropic itself sent a letter to Senators Tim Scott and Elizabeth Warren accusing the Chinese giant Alibaba of trying to illicitly copy its technology through 24,000 fraudulent accounts. Alibaba declined to comment on the matter.
Using data from one system to train another, a process called distillation, is common in AI development. But Anthropic’s and OpenAI’s terms of service prohibit anyone from secretly collecting data for that purpose. It is unclear whether Z.ai used distillation in developing its technology. Either way, distillation alone is not enough to build a cutting-edge AI system, as Charles O Neill, head of model training at Baseten, a company that sells access to GLM-5.2, pointed out. In his view, the narrative that all the capabilities of these models come from Anthropic is not as true as people say.
Chinese models arriving on equal footing
The sudden success of GLM-5.2 is not an isolated case. It is part of a larger trend that has been solidifying over the past several months: Chinese artificial intelligence models are getting closer and closer to the technical level of the best Western models, and in many cases, with significantly lower operating costs. About 18 months ago, the Chinese startup DeepSeek had already shocked Silicon Valley by demonstrating it could build efficient AI in a much more affordable way. Now, Z.ai is repeating the formula with even more precision and perfect timing.
This technical convergence between the Chinese models and those from OpenAI and Anthropic is changing how the market thinks about AI leadership. For a long time, the dominant narrative was that American companies had a technological advantage that was virtually insurmountable, especially because of export controls that limit China’s access to the specialized chips needed to train AI. Z.ai’s corporate documents in Hong Kong show that in the first half of 2025, the company spent more than seven times its revenue on computing service fees, precisely to gain access to those chips outside of China.
Even with those obstacles, experts estimate that China is only six months or less behind American companies. According to Jeffrey Ding, a professor at George Washington University specializing in emerging technologies and international relations, there was speculation that export controls would eventually widen the gap between American and Chinese models, but GLM is pushing things in the opposite direction.
For developers and companies that use artificial intelligence as a core part of their products, this competition is very welcome. More options, with comparable quality and more varied pricing, means more flexibility to build scalable solutions without depending exclusively on a single vendor. And with Fable and Mythos out of circulation, many businesses realized the importance of having alternatives. As Justin Summerville, head of data analysis at OpenRouter, put it, there is a certain apprehension at large organizations about loyalty, after all, nobody knows which model will be the best one three weeks from now.
What is becoming increasingly clear is that the race for artificial intelligence leadership is no longer an exclusively American contest. It is global, it is accelerating, and Chinese models have proven they can compete on equal footing, sometimes even under more favorable cost conditions. For OpenAI, for Anthropic, and for any other company that once felt comfortable at the top of this market, the message is pretty straightforward: the game has changed, and it is more competitive than ever. 🌐
