Grok 4.6 Nearly Matches Claude Fable 5 on Agentic Benchmark at a Fraction of the Cost

ArtificialAnlys · x · 2026-08-13

On Artificial Analysis' AA-Briefcase agentic knowledge work benchmark, xAI's Grok 4.6 made significant gains, scoring neck and neck with Anthropic's Claude Fable 5.

The benchmark evaluates models on long-horizon agentic tasks. Notably, Grok 4.6 delivers a cost per task of just $4.42, substantially cheaper than Claude Fable 5 ($22.30) and Claude Opus 5 ($17.79), demonstrating an impressive intelligence-per-dollar ratio.

Related event: Grok 4.6 Tops Agentic Benchmarks with High Performance and Cost-Efficiency(5 posts)→

Original post →

More from Models

Models channel →