Claude Fable 5.1 tops benchmarks but costs 20% more per task due to higher token usage

ArtificialAnlys · x · 2026-09-02

Artificial Analysis reports that Claude Fable 5.1 (max) scores 66 on the Intelligence Index, surpassing Claude Opus 5 and GPT-5.6 Sol, with new highs on Terminal-Bench v2.1 and SciCode. Despite a 75% cut in cache read pricing, the cost per task is 20% higher than Fable 5 due to increased output tokens. It leads in agentic tasks like GDPval-AA v2 but trails Opus 5 in presentation quality.

Related event: Fable 5.1 Tops Benchmarks, Cost Debate Erupts(14 posts)→

Original post →

More from Models

Models channel →