Omneky Harness Outperforms ChatGPT and Gemini on TasteBench Ad Score

Omneky’s advertising harness scored 8.8 in the TasteBench evaluation, 1.2 points above the next-best tested ChatGPT configuration at 7.6; the test covered 52 ad requests across four brands and 364 ads. On a separate readiness measure, 60% of Omneky ads passed all applicable checks versus 13% for the next-best configuration. The evaluation assessed generated creative, not campaign returns.

Analysis

The useful signal is about workflow integration, not model quality: a proprietary harness can outperform access to frontier models by coordinating brand context, creative direction, and asset selection. That favors application-layer products and raises the bar for generic image-generation APIs, but this small, vendor-run test does not establish durable advantage or better campaign economics. Four brands, 52 requests, and three anonymous judges leave substantial scope for benchmark design and evaluation bias; the readiness gap is potentially more commercially relevant than the taste score, but its checks may not predict advertiser outcomes.

For Alphabet, the read-through is mildly negative for Gemini as a standalone creative proposition, not evidence of material pressure on consolidated results. Alphabet can respond by embedding stronger brand context and workflow controls in Google Ads; distribution and campaign feedback may matter more than a benchmark lead. Adobe and other creative-workflow vendors could also benefit if customers value governance and repeatability over raw model access. The counterpoint: Omneky’s claimed edge may not convert into retention, pricing power, or return on ad spend.

Near term, likely limited market impact given no campaign-return data or independently verified commercial adoption. Over 1–3 months, watch customer wins, repeat usage, and disclosed performance against Google’s native tools. Over 6–18 months, the strategic question is whether workflow vendors capture value or model/platform providers bundle the same capabilities. Falsification: independent, broader tests show no quality/readiness advantage, or advertisers report no improvement in campaign results.

Key Decisions for Investors

• No immediate GOOG trade: this benchmark is too narrow to support a material earnings or valuation revision. Treat the per-company negative sentiment as a watch signal, not a short thesis.
• Monitor Google Ads product releases and advertiser disclosures for brand-context controls, creative workflow adoption, and measurable campaign outcomes; a credible native response would weaken the application-layer threat.
• Track Omneky’s independently verifiable customer adoption, retention, and campaign-return evidence before underwriting a competitive moat. The key missing data are sample representativeness, judge methodology, and results across additional brands and spend levels.
• If independent tests and customer evidence later confirm persistent workflow advantage, revisit relative exposure to standalone creative-generation platforms versus integrated ad platforms; do not infer revenue displacement from TasteBench scores alone.