Sakana's Fugu Ultra v1.1 — AI Model Router Beats Fable 5 Without Even Using It
Japanese AI company Sakana released Fugu Ultra v1.1, a model router that beats Anthropic's Fable 5 across most benchmarks — without even including Fable 5 in its selection pool.
Fugu Ultra v1.1 — The AI Model Router That Beats Fable 5
On July 24, 2026, Sakana AI, the Tokyo-based AI company known for its evolutionary AI approach, released Fugu Ultra v1.1 — an intelligent AI model router that claims to outperform Anthropic's Fable 5 across most benchmarks, without even including Fable 5 in its selection pool.
What is a Model Router?
Unlike traditional AI models that try to handle every task themselves, a model router (like Fugu) intelligently analyzes each incoming request and routes it to the best-suited model for that specific task. This approach can achieve better overall results at lower cost than any single model, even a frontier one like Fable 5.
Fugu Ultra v1.1 — What's New?
Sakana claims performance gains of up to 7.9 points over v1.0, with the biggest jumps on ProgramBench and TerminalBench 2.1. The router reportedly beats Fable 5 across most benchmarks, even though Fable 5 isn't part of the router's selection pool — meaning it achieves this without direct access to Anthropic's best model.
Pricing and Availability
Sakana has published pricing and a technical report for Fugu Ultra v1.1. Notably, the company also offers a Claude Code-compatible endpoint, making it easy for developers to integrate Fugu into their existing workflows.
Why Model Routing Matters
Model routing represents a fundamentally different approach to AI optimization. Instead of building one bigger model, route tasks intelligently to the best model for each use case. This could mean lower costs, better results, and more flexibility for businesses and developers building on AI.
The Catch: Self-Reported Results
All of these numbers come from Sakana itself, and no independent verification exists yet. The first Fugu version received a lukewarm reception, so the v1.1 claims should be taken with appropriate skepticism until third-party validation emerges.
Conclusion
Fugu Ultra v1.1 represents an innovative approach to AI optimization. Instead of just building bigger models, Sakana is building smarter systems that route tasks optimally. Whether the v1.1 claims hold up to independent verification remains to be seen, but the direction is undeniably interesting.