• The ban that opened a market
• China and Japan move in
• Competing narratives on dependence and control
• A shifting AI landscape

Asian AI Startups Launch Alternatives to Anthropic’s Restricted Models Asian AI startups are racing to fill a gap created by Washington’s export controls, launching homegrown systems they say can match Anthropic’s most powerful, now-restricted models — and reshaping the global AI map in the process.

§The ban that opened a market

Two weeks ago, the U.S. government barred non-Americans from accessing Anthropic’s cybersecurity-focused Mythos model and its more restricted sibling Fable 5, citing their exceptional power and potential misuse risks. As the order entered its third week with no resolution, access to some of the world’s most advanced commercial AI suddenly became uncertain across Asia.

§China and Japan move in

Earlier this week in Tokyo, Sakana AI launched Fugu, a seven‑billion‑parameter “orchestrator” model designed to route tasks among multiple external systems. The company says Fugu “stands shoulder-to-shoulder with leading models like Anthropic’s Fable 5 and Mythos Preview,” matching their performance by coordinating cheaper models through APIs rather than training a single massive system. Sakana, founded in 2023 by former Google researchers and a former Japanese diplomat, has raised $135 million and showcased its orchestration research at ICLR this spring.

In Beijing, cybersecurity firm 360 Security unveiled Tulongfeng at the ISC AI 2026 conference, pitching it as a vulnerability‑discovery tool able to “go head‑to‑head” with Mythos. Founder Zhou Hongyi framed such capability as a national strategic asset.

§Competing narratives on dependence and control

Sakana insists the launch timing was “entirely coincidental,” stressing Fugu had been in development since last year, but its website explicitly touts “delivering frontier capability without the risk of export controls.” The firm argues its main customers — Japanese businesses and agencies — want to reduce exposure to tightening U.S. rules, even as it concedes “U.S. models remain important to Asia.”

Critics of Washington’s strategy warn that bans may simply accelerate non‑U.S. innovation. One AI researcher argued that restricting open models for U.S. companies “won’t stop … global open model progress” or “bad actors using them,” questioning “what exactly is gained by banning open models, including those from China?”

§A shifting AI landscape

Together, Fugu and Tulongfeng signal a broader push by Asian firms to build high‑end AI that cannot be switched off by a foreign government. Whether they fully rival Anthropic’s systems or not, the export ban has already achieved one clear effect: it has made the risk of over‑reliance on U.S. AI impossible for Asia’s tech sector — and its competitors — to ignore.

Continue reading https://foxvector.com/stories/019f0b67-0e15-2056-7200-3e8da0687fff