Orchestrator Sakana Fugu Claims Fable 5 Parity: Real-World Tests Reveal 30-Minute Waits
Orchestrator Sakana Fugu claims parity with Fable 5, but real-world testing has highlighted longer waits and weaker coding outcomes since launch. Tokyo-based Sakana AI rolled out Fugu on June 22, nearly ten days after the U.S. Department of Commerce suspended international access to Anthropic’s Fable 5 and Mythos Preview under an export control directive. In the first 24 hours, developers reported about 30-minute wait times on Fugu Ultra and results that fell short of the Fable 5 standard, contrasting with vendor-reported benchmarks. Fugu, launched as a single OpenAI-compatible API, routes requests across an internal pool of specialist models. Sakana says its routing intelligence is embedded in trained weights and uses recursive self-calling, based on ICLR 2026 papers describing a reinforcement-trained Conductor and a coordination approach called TRINITY.







