A benchmarking prompt used to compare the realism of different AI models (including GPT Image 2.5) f