Omneky Launches TASTE BENCH to Measure AI Creative Readiness

The TASTE BENCH leaderboard evaluates output based on creative execution, brand alignment, and technical viability. By removing human oversight from the initial generation phase, the system provides a benchmark for which models can reliably produce professional-grade ads without manual correction. Models are judged on eleven strict pass-fail criteria, including typography accuracy, logo fidelity, and the absence of visual artifacts.

"Taste is in the details: whether the typography feels right, whether the product looks authentic, and whether the idea comes through immediately," said Hikari Senju, CEO of Omneky. To achieve these insights, the company employed a blind panel of AI judges from Google, Anthropic, and OpenAI to score 469 generated ads. While the current iteration focuses on image models, the data highlights significant performance gaps, with readiness rates for the tested models ranging from 22% to 54.2%. These metrics focus on the immediate viability of the asset rather than downstream campaign conversions.