Sakana says Fugu-Ultra v1.1 adds up to 7.9 points on coding and agentic benchmarks

SakanaAILabs · x · 2026-07-24

Sakana releases Fugu-Ultra v1.1 with stronger coding and agentic results

Sakana AI announces Fugu-Ultra v1.1, an updated model in the Fugu family. The company says the new version incorporates the latest frontier models and improves performance across every benchmark shown, with gains of up to 7.9 points over v1.0.

Notable improvements were reported on ProgramBench and Terminal Bench 2.1. Sakana says the model is more capable on:

Pricing remains unchanged versus Fugu-Ultra v1.0.

Related event: Sakana AI Introduces Fugu-Ultra v1.1 with Enhanced Coding Capabilities(4 posts)→

Original post →

More from coding & agent

coding & agent channel →