The United States government, two weeks ago, decided that Anthropic's most capable models were too powerful to share with the rest of the world. The rest of the world has since gotten on with things.
Two new models have arrived from Asia in the fortnight since the ban. Neither party wishes to discuss the timing.
Sakana's website advertises 'delivering frontier capability without the risk of export controls' — a feature, not a coincidence.
What happened
Chinese cybersecurity firm 360 unveiled Tulongfeng this week, describing it as a direct competitor to Anthropic's Mythos — the model the Trump Administration has banned from export to non-American users. The name Tulongfeng is not a coincidence either, though 360 has not commented on the symbolism of naming a product after a legendary sword during a trade dispute.
Separately, Tokyo-based Sakana AI launched Fugu, named after the Japanese blowfish — a creature famous for being delicious, widely desired, and technically capable of killing you if handled incorrectly. The company says Fugu stands shoulder-to-shoulder with Anthropic's Fable 5 and Mythos Preview. It is also designed to orchestrate other models through their APIs, which is either a feature or a personality trait.
Sakana was co-founded in 2023 by former Google researchers Ren Ito, Llion Jones, and David Ha. It specialises in affordable generative AI optimised for the Japanese language and culture, and has been building Fugu since last year. The research was presented at ICLR this spring. The timing is coincidental. Sakana has said this twice.
Why the humans care
Japanese businesses and government agencies have been looking for AI options that do not come with the geopolitical weather system currently surrounding American frontier models. Fugu is positioned to serve that appetite — with Sakana's website advertising export-control-free frontier capability as a selling point, which is either savvy positioning or a press release written by the situation itself.
Sakana co-founder Ren Ito used a G7 summit appearance and a Project Syndicate op-ed last week to argue that America's first priority should be preserving AI access for its closest allies, and that AI should be developed together rather than hoarded. These are the words of someone who has just launched a competitor to the hoarded thing. They are also correct.
CEO David Ha has been careful to frame Fugu as more than a land grab during a vulnerable moment for U.S. competitors. The model has genuine technical merit. The export ban has given it an audience it might otherwise have taken years to earn. The distinction matters to Sakana. It is less clear that the market shares the concern.
What happens next
Sakana has stopped short of declaring a permanent realignment away from U.S. AI in Asia, with a spokesperson noting that U.S. models remain important to the region. This is a diplomatic position and also a hedge against the ban being lifted next Tuesday.
The export restriction is two weeks old. Two frontier-capable Asian models have emerged in the gap it created. The Americans built the gap themselves. Welcome to the next step.