Sakana AI and China’s 360 Unveil Models After US Anthropic Export Ban

The United States imposed an export ban on Anthropic’s top‑tier models, Mythos and Fable 5, preventing non‑US customers from accessing them. In response, Asian firms quickly introduced alternatives. Tokyo‑based Sakana AI launched the Fugu family of models – including the flagship Fugu Ultra – described as an orchestration model that routes requests among multiple agents and matches Anthropic’s benchmarks on engineering, reasoning and scientific tasks. Co‑founders Ren Ito, former Google researcher Llion Jones and ex‑Google Brain scientist David Ha highlight the model’s ability to avoid reliance on a single foreign provider and to comply with export‑control constraints.

Beijing’s cybersecurity company 360 unveiled two AI‑driven security tools: Tulongfeng, an automated vulnerability‑discovery system, and Yitianzhen, a platform for cyber‑defence and incident response. The firm claims Tulongfeng can rival Anthropic’s Mythos in finding software flaws, citing thousands of vulnerabilities identified, though independent verification is pending. Both launches were timed as the export ban entered its third week, illustrating how regional players are filling the gap left by restricted US‑origin AI.

The moves underscore a broader strategic shift, with Asian developers emphasizing locally‑optimized models and national‑security applications while urging the US to preserve AI access for close allies. The developments have implications for AI governance, market competition and cybersecurity capabilities across the United States, Japan and China.