Asian AI startups launch Mythos-like models as Anthropic’s export ban drags on

On Wednesday, 360, a Chinese cybersecurity firm, launched Tulongfeng, an AI tool touted as a competitor to Anthropic’s Mythos. This cybersecurity-centric model has garnered attention due to the Trump Administration's export ban on Mythos and its restricted variant, Fable 5, which limits access for non-U.S. citizens. Earlier in the same week, Tokyo-based Sakana AI introduced Fugu, named after the Japanese term for blowfish. The company claims that Fugu stands alongside leading models like Anthropic’s Fable 5 and Mythos Preview, and is specifically designed for agents, enabling them to interact with other models via their APIs. These developments come in the wake of the U.S. government's restrictions, which were enacted two weeks prior, preventing Anthropic from providing global access to its advanced models. A representative from Sakana AI emphasized that the timing of Fugu's release was coincidental, yet the company is keen to leverage the current market dynamics. Their website highlights a commitment to delivering cutting-edge capabilities without the constraints of export regulations. The spokesperson remarked, "Sakana Fugu is something we have been building since last year — the research behind it was presented at ICLR this spring. We were confident in the product on its own merits; the timing simply happened to coincide with a moment that brought it more attention than we expected." Founded in 2023 by ex-Google researchers Ren Ito, Llion Jones, and David Ha, Sakana aims to create affordable generative AI models optimized for Japanese language and culture, targeting local businesses and government agencies. Although the company is positioning Fugu as a solution to mitigate risks from tightening export controls, they acknowledge the continued relevance of U.S. AI technologies in Asia. "U.S. models remain important to Asia," the spokesperson stated, echoing sentiments expressed by co-founder Ren Ito at the recent G7 summit regarding AI access and export regulations. In an op-ed published by Project Syndicate, Ito urged the U.S. government to prioritize maintaining access for its closest allies and argued against hoarding AI technology, advocating for collaborative development instead. David Ha, Sakana’s co-founder and CEO, described Fugu not merely as a strategic move amid competitor vulnerabilities but as a tool designed to facilitate coordinated agent usage across multiple models. He expressed concerns about depending on a single provider for national infrastructure, especially in light of recent export controls. "Access to top models can disappear overnight," he warned. "Collective intelligence is the practical hedge against this concentration of power." While Sakana seeks to position Fugu as a safeguard for AI access rather than a complete replacement for U.S. technologies, 360’s approach appears more aggressive. The Chinese firm recently unveiled two AI security tools: Tulongfeng for identifying software vulnerabilities and Yitianzhen, aimed at automating cyber defense and incident response. 360's founder, Zhou Hongyi, emphasized the strategic importance of vulnerability-detection AI and highlighted the dangers posed by unequal access to advanced tools. Despite Anthropic's impressive growth, with revenue projections exceeding $47 billion by May 2026, the exact dependency on Asian enterprise customers remains unclear. Since the export restrictions were imposed, at least two companies, one in Tokyo and another in Beijing, have stepped into the vacated space, potentially altering the competitive landscape. Even if U.S. firms regain trust in the future, local alternatives are already emerging, tailored to better understand regional languages and cultural nuances. 360 did not respond to requests for further comment.