Fable 5.1 doubles scientific benchmark score, surpassing Opus 5

felixrieseberg · x · 2026-09-02

On the Stanford-led Terminal-Bench-Science, Fable 5.1 scored 52.6%, more than doubling Fable 5 (24.7%) and beating Opus 5 (29.0%). This marks significant progress in AI scientific reasoning capabilities.

Related event: Fable 5.1 Doubles Science Benchmark Score and Refines Writing Style(3 posts)→

Original post →

More from Models

Models channel →