tech
Asian AI startups launch Mythos-like models as Anthropic's export ban drags on
Sakana AI and China’s 360 Security launched AI tools as alternatives to Anthropic’s banned Mythos and Fable models.
TL;DR
- Sakana AI launched Fugu, an orchestration model that matches Fable 5 performance.
- 360 Security unveiled Tulongfeng, a vulnerability-discovery tool comparable to Mythos.
- These launches occurred during the third week of the US government's export ban on Anthropic's AI models.
- Sakana AI's Fugu uses an orchestration approach, routing tasks across a pool of models.
- 360 Security's Tulongfeng targets automated vulnerability discovery and is considered a national strategic asset in China.
- Zhou Hongyi of 360 Security acknowledged a gap in capability between Chinese and US models but stressed the need for independent development.
- The trend reflects a broader Asian movement towards sovereign AI development, with India planning a $5 billion fund.
- Ren Ito of Sakana AI advocated for shared AI development rather than hoarding, suggesting a shift in perspective for frontier models.
- Neither company claims to replace American frontier AI entirely, but rather to offer resilient alternatives.
- The commercial reality is seeing competitors emerge rapidly due to export restrictions, with dependence on American AI now seen as a significant risk.