Asian AI startups launch Mythos-like models

	Asian AI startups launch Mythos-like models(techcrunch.com)
	250 points by bogdiyan 23 hours ago \| 181 comments
	tl;dr: China's 360 unveiled Tulongfeng, a cybersecurity-focused AI model pitched as a rival to Anthropic's Mythos, while Tokyo-based Sakana AI launched Fugu, an orchestration model designed to coordinate access to multiple models via APIs. Both launches follow a US export ban two weeks ago blocking non-Americans from accessing Anthropic's Mythos and Fable 5. Sakana frames Fugu as a hedge against US access risks rather than a replacement, while 360's founder cast vulnerability-finding AI as a national strategic asset.
	HN Discussion: ↓Fugu underperforms Opus while being slow and expensive in real-world testing •Fugu is an orchestration system routing to other models, explaining the high costs from double API markup ↓Calling these models 'Mythos-like' is meaningless without reliable benchmarks for comparison ~Expect US to ban foreign LLMs under vague safety pretexts regardless of actual performance ~Mythos being unavailable conveniently makes the 'Mythos-like' claim unfalsifiable