Sakana AI and 360 launched new AI models this week as a U.S. export order restricting Anthropic’s Mythos and Fable to American users continues to block global access.
The order, issued two weeks ago, prevents Anthropic from making Mythos and the more restricted Fable 5 available to non-Americans, creating a market opening for regional alternatives in Asia.
"Orchestration Models are the next frontier, beyond bigger models," Sakana co-founder and CEO David Ha wrote on X.
Sakana said its Fugu model, named after the Japanese blowfish, is designed for agent use, can orchestrate access to other models via APIs and matches what the company describes as frontier capability.
Sakana, cofounded in 2023 by former Google researchers Ren Ito, Llion Jones and David Ha, builds compact generative models tuned for Japanese language and culture and is targeting businesses and government clients wary of export controls.
The company said it began work on Fugu last year, presented related research at ICLR this spring, and framed the launch as product-driven rather than a direct reaction to Anthropic’s export restrictions.
Beijing-based cybersecurity firm 360 unveiled Tulongfeng for automated vulnerability discovery and Yitianzhen for automated defence and incident response, with founder Zhou Hongyi calling vulnerability-detection AI a national strategic asset and warning that unequal access creates risk.
Anthropic reported a run-rate revenue of $47 billion in May 2026, underscoring the scale of the capability now partly restricted by the export order.
Even if U.S. companies regain global access, Sakana and 360’s launches show local alternatives tuned for language and regulatory nuance are already filling the gap.
