Issue 01 / Marketing Taste Tests

Compare AI models on real briefs.

Three quick blind campaigns. Pick the creative output you would ship, reveal the models behind them, and see what recorded voters picked.

Read. Pick. Reveal.

Read both anonymous outputs and click on the preferred one to reveal model details and crowd ratios.

Ten blind duels. Select the better AI output to cast your vote.
Duel 1 of 10Brand VoiceTested 2026-07-06

Prompt used

Write a one-line brand voice sample for a premium sparkling water called Lumen. It should feel crisp, adult, and quietly luxurious without sounding like perfume copy.

Editorial desk with anonymous A and B campaign tear sheets, color swatches, and review notes
Hero Illustration
Option A
GPT-5.6 Sol (OpenAI)
--%voters

Lumen is sparkling water with a cleaner glow: light citrus, fine bubbles, and nothing trying too hard.

Pros

  • Very clear
  • Clean benefits list
  • Low risk

Cons

  • Slightly generic product words
  • Less emotional grip

Best for

High-conversion copyComplex campaign logicLifecycle messaging
Scoring note: Clean, but a bit safe. Good for high-volume consumer appeal.
Option B
Claude Sonnet 5 (Anthropic)
--%voters

Lumen is a bright, mineral snap of sparkling water: clean enough for noon, elegant enough for after dark.

Pros

  • Sharp sensory descriptors
  • Strong brand lifestyle bridge
  • Distinct tone

Cons

  • “Mineral snap” could split consumer taste
  • Slightly complex structure

Best for

Brand voice & toneSensory ad hooksLong-form editorial strategy
Scoring note: Builds a clear evening-use occasions hook for premium buyers.

The rule

No fake crowd scores.

Your vote stays local unless you share a challenge. Each verdict is for the tested task on the tested date, not a claim about the best AI overall.

Read the method