Sakana AI launches Fugu Ultra and its orchestration model matches Fable and Mythos without training a single frontier model

Sakana AI launched Fugu and Fugu Ultra in Tokyo today, a multi-agent orchestration system that scores 54.2 on SWE-Pro and 95.1 on GPQA-Diamond by routing tasks across a pool of frontier models rather than training one. The approach matches Anthropic's Fable and Mythos on key benchmarks while eliminating export control risk, with pricing starting at $5 per million input tokens.