Benchmark Prompts / P12

Chess set: 7 AI models, one prompt

Every Season 1 competitor received this prompt verbatim, at default settings, with one generation and no manual cleanup. Nothing below is cherry-picked — the failures ship too, because a benchmark that hides them isn't one.

The prompt (v1.0, CC BY 4.0)

A chess set arrangement: checkered board with the two front rows of white pieces placed correctly.

What it tests: Multi-object precision, placement rules, and instruction following.

Category: Abstract

Blind vote results for P12

93 ballots · 6 “both bad”

CompetitorWinsLosses
GPT-5.6 Sol162
Claude Opus 4.81416
Claude Fable 51214
Claude Opus 5121
Grok 4.6123
Claude Sonnet 5512
Claude Haiku 4.5225

Every output for this prompt

Click any card to load the model in 3D and orbit it. Generation notes under each entry state exactly how it was produced.

P12: “A chess set arrangement: checkered board with the two front rows of white pieces placed correctly.

P12: “A chess set arrangement: checkered board with the two front rows of white pieces placed correctly.

P12: “A chess set arrangement: checkered board with the two front rows of white pieces placed correctly.

P12: “A chess set arrangement: checkered board with the two front rows of white pieces placed correctly.

P12: “A chess set arrangement: checkered board with the two front rows of white pieces placed correctly.

GPT-5.6 SolAbstract

P12: “A chess set arrangement: checkered board with the two front rows of white pieces placed correctly.

Grok 4.6Abstract

P12: “A chess set arrangement: checkered board with the two front rows of white pieces placed correctly.