Sakana AI has released Fugu Ultra v1.1, an updated AI router that routes each query across a pool of publicly available top-tier models. The company claims performance improvements of up to 7.9 points over the previous version, with the most significant gains on ProgramBench and TerminalBench 2.1. Notably, Fugu Ultra v1.1 reportedly outperforms Anthropic's Fable 5 across most benchmarks, even though Fable 5 is not part of the router's selection pool. All performance metrics are attributed to Sakana itself, with no independent verification currently available. Pricing for the service remains at $5 per million input tokens and $30 per million output tokens. Source: thedecoder
The update also includes a Claude Code-compatible endpoint, allowing users to call Fugu directly from the terminal. Sakana said it takes about two weeks of training and evaluation before a new top-tier model is added to the pool. The architecture is detailed in the technical report, though the company has not provided specific details about the model's design or training process. Since its launch, Fugu has been available on platforms like OpenRouter and Vercel. However, the first version of Fugu received mixed reviews, with critics citing high token usage, slow response times, and subpar results. Source: thedecoder
Sakana AI still does not serve the EU or EEA, citing GDPR and "EU-specific regulations" as the reason. The company has not provided further details about its expansion plans or regulatory compliance efforts. Source: thedecoder
