Trending storyDeveloping

Artificial Analysis ranks Grok 4.5 to 4.7 higher on AA-PersonalityBench

1 article1 sourcesince Oct 29Last article 2h ago · Oct 29, 5:48 PM ET

Overview

AISummary of 1 article

Artificial Analysis reports that Grok 4.5, 4.6 and 4.7 improved on its AA-PersonalityBench across three measures: slop, sycophancy and sincerity.

Grok 4.7 ranks 6th percentile on Slop, 1st on Sycophantic and 90th on Sincere. From Grok 4.5 to 4.7, the Slop percentile fell 62 points, the Sycophantic percentile fell 17 and the Sincere percentile rose 64. Grok 4.5 scored 68 on Slop, 18 on Sycophantic and 26 on Sincere.

The source reports these scores as percentile ranks, and the post does not define how the benchmark scores each measure beyond these figures.

Written by AI from the articles below · updated Oct 29, 5:58 PM ET

Check the sources:

Latest developmentOct 29, 5:48 PM ET

Grok 4.5 to 4.7 get less sycophantic, less slop, more sincere

Article timeline

The articles in this story. Times are ET.

Oct 29

5:48 PM
Artificial Analysis@ArtificialAnlysOfficial
Grok 4.5 to 4.7 get less sycophantic, less slop, more sincere
AIArtificial Analysis reports that the last three Grok releases improved on its AA-PersonalityBench, with Grok 4.7 ranking 6th percentile on Slop, 1st on Sycophantic, and 90th on Sincere. Across Grok 4.5 to Grok 4.7, Slop fell 62 percentile points, Sycophantic fell 17, and Sincere rose 64. Grok 4.5 scored 68, 18, and 26 on Slop, Sycophantic, and Sincere, respectively.

Heat Trend

Not enough continuous observations to show a trend yet.