Vaibhav Sisinty@VaibhavSisintyx73.7 SWE-Bench Pro from Fugu Ultra routing multiple frontier models
- Sakana Fugu is a coordinator that routes API calls to the best frontier model for each subproblem.
- Fugu Ultra scores 73.7 on SWE-Bench Pro, beating Claude Opus 4.8 (69.2) and GPT-5.5 (58.6).
- The model pool is swappable, allowing routing around export-controlled providers.
how to reproduce · Send one API call to Fugu; it decides which frontier model handles each part of the problem, runs them together, and returns a single answer.