Chinese and Japanese AI Startups Roll Out Mythos-Like Models While Anthropic Likely to Lose Out

Thailand AI News: Asian AI Firms Move Quickly as Export Ban by Anthropic Continues

Asian artificial intelligence companies are wasting little time capitalizing on the uncertainty created by the United States’ export restrictions on Anthropic’s most advanced AI models. New AI offerings from Japan’s Sakana AI and China’s cybersecurity firm 360 Security are emerging as credible alternatives for businesses and government agencies seeking frontier AI capabilities while access to Anthropic’s Mythos and Fable 5 models remains restricted.

The U.S. government’s export controls have now entered their third week with no clear indication of when they may be lifted. The restrictions have forced many organizations across Asia to reconsider their dependence on a single American AI supplier, creating opportunities for regional developers to showcase home-grown technologies capable of meeting enterprise and public-sector requirements.

Sakana AI Unveils Fugu

Tokyo-based Sakana AI has introduced Fugu, a frontier orchestration model named after the Japanese blowfish. Unlike conventional large language models that attempt to solve every task independently, Fugu coordinates multiple AI models through their APIs, intelligently assigning different parts of a problem to whichever model is best equipped to handle them.

According to Sakana AI, this orchestration approach enables Fugu to achieve performance comparable to Anthropic’s Fable 5 and Mythos Preview while requiring only a fraction of the computing resources needed to build a new frontier model from the ground up.

The company told Thailand AI News that Fugu is designed primarily for Japanese businesses and government agencies looking to reduce the risks associated with export controls and supply uncertainty. Promotional material for the platform highlights its ability to deliver frontier AI capabilities without depending on technologies affected by international trade restrictions.

Timing Coincides with Anthropic Restrictions

Although Fugu was launched shortly after Anthropic’s export restrictions took effect, Sakana AI says the timing was purely coincidental. Company representatives explained that development began last year and that the research behind the model was presented earlier this year during the International Conference on Learning Representations (ICLR).

The company said it had complete confidence in the technology based on its own research achievements and engineering work, adding that the export restrictions simply generated far greater interest in the launch than originally expected.

Founded in 2023, Sakana AI was established by former Google researchers David Ha and Llion Jones together with former Mercari and Stability AI executive Ren Ito. The startup has focused on building efficient generative AI models optimized for Japanese language, culture and business applications while minimizing computational costs.

A Different Vision for Frontier AI

David Ha believes orchestration models represent the next major evolution of artificial intelligence. Rather than building increasingly larger standalone models, he argues that coordinating several specialized AI systems can deliver similar or even better results while reducing costs and minimizing reliance on a single provider.

Ha also warned that the recent export restrictions demonstrate how quickly access to leading AI technologies can disappear due to geopolitical decisions. In his view, organizations need diversified AI strategies that reduce exposure to sudden policy changes.

Ren Ito has also stressed that the company is not attempting to replace American AI altogether. Instead, Sakana sees Fugu as a hedge against uncertainty, maintaining that U.S. AI companies remain valuable partners for Asia while encouraging greater regional resilience.

In a recent opinion article published by Project Syndicate, Ito called on the United States to preserve access to advanced AI technologies for allied nations, arguing that frontier AI should be developed collaboratively rather than becoming a tightly controlled strategic resource.

China’s 360 Security Enters the Race

While Sakana AI has positioned Fugu as an orchestration platform, China’s 360 Security has taken direct aim at AI-powered cybersecurity.

During the ISC AI 2026 cybersecurity conference in Beijing, founder Zhou Hongyi unveiled two new products under the Yitian Tulong platform. Tulongfeng is designed to automatically discover software vulnerabilities, while Yitianzhen focuses on cyber defence, threat detection and automated incident response.

Zhou described vulnerability-discovery AI as a national strategic capability and warned against what he called “one-way transparency,” where only a limited number of countries possess sophisticated vulnerability detection technology while others remain unable to identify weaknesses within their own software infrastructure.

Although Zhou acknowledged that Chinese AI models continue to trail leading American systems by approximately 20 to 30 percent in core capabilities, he argued that waiting for complete technological parity would only slow China’s cybersecurity progress.

According to the company, Tulongfeng has already identified more than 3,400 software vulnerabilities, including over 100 reportedly confirmed by Chinese authorities, although those figures have not been independently verified.

Asia’s AI Landscape Continues to Shift

The launches by Sakana AI and 360 Security reflect a broader transformation taking place across Asia’s AI industry. Governments and businesses are increasingly investing in regional AI ecosystems to reduce dependence on foreign providers and strengthen technological resilience amid growing geopolitical uncertainty.

The debate over AI sovereignty has also intensified. India is considering a multi-billion-dollar sovereign AI fund, while AI access and export controls featured prominently during discussions at the recent G7 summit, highlighting the strategic importance of frontier AI technologies.

Anthropic continues to report impressive commercial growth, with annualized revenue exceeding US$47 billion in May 2026. However, the company has not disclosed how much of its enterprise business comes from Asian customers, leaving unanswered questions about the long-term financial impact of the export restrictions.

Regional Competition Is Accelerating

Whether the export restrictions are eventually lifted or remain in place, the competitive landscape has already begun to change. Within weeks of the U.S. decision, companies in both Japan and China introduced advanced AI platforms positioned to serve markets that previously depended heavily on American frontier models. As these alternatives continue to mature, they could reshape Asia’s AI ecosystem and gradually erode Anthropic’s competitive advantage across one of the world’s fastest-growing artificial intelligence markets.

For more on Sakana AI’s Fugu Model, visit: https://sakana.ai/fugu-release/

For more on 360’s Tulongfeng model, visit: https://www.360.cn/

For the latest developments in the global AI industry, keep on logging to Thailand AI News.