Stop Slop StudioThe Stop Slop Benchmark
Which AI writes fiction that sounds human? Each score combines five assessments: ideas, outlines, a short story, a novel opening, and a scene in another genre.
One page of our rules made 21 of 21 sound more human.
Tap any model to see its score on every task and what the judges wrote.
Total score out of 80: five quality scores plus three sounds-human scores, each 1 to 10. The tabs cycle on their own; tap one to stop on it. Tap a model for its breakdown.
How we tested
Five tasks for every model: fantasy ideas, outlines, a full short story, a novel opening scene, and a non-fantasy mystery scene with a romance in it.
Round one, each model wrote with its default settings and no help. Round two, the exact same tasks with one page of the Stop Slop method placed first, the same page for every model.