US AI export controls have reshaped global access to frontier AI in a matter of weeks, and the ripple effects are already reaching developers and businesses in Pakistan. In mid-June 2026, the Trump administration ordered Anthropic to block all foreign nationals from using its two newest models. The administration of US President Donald Trump barred foreigners from accessing the top AI models developed by Anthropic, citing national security concerns, less than a week after the company rolled out new models named Claude Fable 5 and Mythos 5. The move sent shockwaves across the global AI industry and, almost immediately, opened a door for China’s Zhipu AI.
What the US AI Export Controls Actually Did
The US government, citing national security authorities, issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. That is a sweeping restriction. It did not just target rival countries. It cut off allied nations, international researchers, and even Anthropic staff born abroad.
Anthropic said: “The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.” Access to all other Anthropic models was not affected, and the latest measures aim to decide who can use software developed in the US as part of the Trump administration’s policy of export control over high technology, especially in critical sectors such as AI and semiconductors.
The ban was then partially lifted by the Commerce Department, with the US government asserting a new level of influence over AI by controlling which companies can access Anthropic’s new models, while OpenAI agreed to let the administration screen users of its new model. In short, even after the partial reversal, Washington now decides who gets access to the most powerful American AI tools.
The order is part of the Trump administration’s broader policy of export controls over advanced technology, the same approach already applied to AI chips from companies like Nvidia and AMD. What is new is extending that logic from hardware to the AI models themselves.
Why Zhipu AI Moved Fast to Fill the Gap
While Washington was arguing over the ban, Beijing-based Zhipu AI was making noise. Shares in Chinese lab Zhipu (also known as Z.ai) jumped more than 30% after it released a new open-source model, with the company arguing pointedly that frontier intelligence should not be subject to withdrawal at the stroke of a rule.
As China’s first AI software maker to list publicly, Beijing-based Zhipu AI made waves on January 8, 2026, with its Hong Kong IPO and a bold prediction about the future of global AI competition. Zhipu’s defining 2026 corporate event was its Hong Kong IPO on 8 January 2026 (stock code 2513.HK), which it billed as the world’s first listing by a large-language-model company. The offer raised about HK$4.17 billion and was retail-oversubscribed by more than 1,000 times, with around 70% of proceeds earmarked for model R&D through 2028.
The company’s technical progress is real. Zhipu’s GLM-5 delivers enhanced coding capabilities and provides users with the ability to perform long-running agentic tasks, and it can directly compete with Anthropic’s Claude Opus 4.5 in coding benchmark tests while surpassing Google’s Gemini 3 Pro in certain benchmarks. Critically, the model was developed using domestically manufactured chips, including Huawei’s Ascend chip, underlining China’s recent successes in building domestic chip self-sufficiency. For developers worried about supply-chain politics, that last point matters a lot.
The Price Gap That Changes Everything for Pakistan
Here is where Pakistani developers and startups need to pay close attention. The price difference between US and Chinese AI tools is enormous. Zhipu’s AI coding tool costs as little as 20 yuan (less than $3) per month, roughly one-seventh the price of Anthropic’s Claude. By late 2025, this low-cost offering had attracted 150,000 paying developers across 184 countries and hit over 100 million yuan ($13.9 million) in annual recurring revenue.
For a Pakistani startup or freelancer working in rupees, that price gap is huge. US AI export controls have now made the higher-priced US tools harder to access too, which pushes cost-sensitive markets like Pakistan further toward Chinese alternatives. By offering lower costs, Zhipu can attract developers and enterprises that might otherwise rely on US platforms, particularly in emerging markets where cost sensitivity is high.
Chinese lab Z.ai’s shares jumped over 30% after a new open-source release, DeepSeek closed a record funding round of about $7.4 billion, and several Chinese labs cut prices sharply, with demand for Chinese models reportedly overtaking US models on some access platforms.
You can learn more about the financial pressure behind US AI labs in our earlier piece on Anthropic’s near-trillion-dollar IPO valuation, which gives useful context on why these companies are under such heavy scrutiny.
Zhipu’s Push into Emerging Markets
Zhipu is not just waiting for US restrictions to send users its way. It is actively building a global network. Zhipu AI’s global expansion is deeply intertwined with China’s Belt and Road Initiative, under which it has established innovation centers in Indonesia, Vietnam, and Malaysia. These hubs offer ‘sovereign AI’ solutions, emphasizing data localization and audit-readiness, to governments and enterprises in emerging markets.
Zhipu leads the ‘International Alliance for Independent Large Model Co-construction,’ partnering with 10 ASEAN countries and 10 Belt and Road nations to build controlled national AI infrastructure, making it a key player in ‘sovereign AI competition,’ a status recognized by OpenAI, which named Zhipu a major rival in its 2025 report on Chinese AI progress. Pakistan, as a country with Belt and Road ties, sits in exactly the market segment Zhipu is targeting.
As a China-based provider, hosted GLM models align with PRC content rules, a consideration for politically sensitive or neutral use. This is something Pakistani businesses and developers should factor in when choosing a platform. Open-source models that you run yourself avoid this issue, but hosted API access is a different matter.
What This Means for Pakistan’s Developers and Businesses
Pakistani software houses, freelancers, and AI startups now face a clearer but trickier choice. US AI export controls have created real uncertainty around American tools. Artificial intelligence has become the latest issue to drive a wedge between the United States and its allies, after President Trump ordered Anthropic to cut off foreign access to its most powerful AI models, citing national security concerns. Pakistan is not a US ally in the formal security sense, so it has even less guaranteed access to protected American AI tools going forward.
On the other side, China’s attractive open-source frameworks and low adoption costs could contribute to a significant competitive advantage over international rivals, which is good news for cost-sensitive users in Pakistan. Zhipu’s open-weight models on Hugging Face can be downloaded and run locally, which removes both the price barrier and the content-filtering concern. Still, data security, sovereignty, and long-term reliability are all factors to weigh.
The broader takeaway for Pakistan’s tech community is this: the global AI platform market is no longer a US monopoly. US AI export controls are accelerating a split, and Pakistani developers should track both sides of the divide rather than assuming US tools will always be available or affordable. If you are building AI-powered products, having a backup plan that includes open-source Chinese models is now practical risk management, not just a budget trick. See also our piece on how the AI memory shortage is already pushing up prices for Pakistani users.
Frequently Asked Questions
What exactly did the US do to Anthropic’s AI models?
On 12 June, the Trump administration issued an export control directive to suspend all access by any foreign national to Anthropic’s newly released Fable 5 and Mythos 5 AI systems, citing national security concerns. The directive applied to any foreign national inside or outside the US, including Anthropic employees who are foreign nationals.
Did OpenAI face the same restrictions?
Anthropic’s rival OpenAI said it had agreed to let the administration screen users of its newest model. So while OpenAI was not hit with a hard ban, it has handed the government control over who can access its most powerful tools. The direction of US AI export controls applies pressure to both companies.
What is Zhipu AI and why does it matter to Pakistan?
Zhipu AI was founded in 2019 as a commercial spin-out of Tsinghua University’s Knowledge Engineering Group by professors Tang Jie and Li Juanzi. Its academic heritage runs through the GLM (General Language Model) architecture and the open ChatGLM dialogue models that made it well known in 2023. Its models are open-source, cheap, and available globally, which makes them a real option for Pakistani developers priced out of or blocked from American tools.
Are there risks in switching to Chinese AI models?
Yes. Starting in 2025, Zhipu has been added to the US Department of Commerce’s Entity List (trade blacklist), amid US concerns over Chinese AI firms and national security. That means using Zhipu’s services could create complications if your business also works with US partners or US government-regulated clients. You should check with a legal adviser if your contracts have US technology compliance clauses. The open-source models hosted locally are generally lower risk than using Zhipu’s commercial API directly.
