Methodology

How we test creative-AI tools

No vibes, no vendor decks. One identical brief, every tool, a fixed rubric, and the real outputs on the page. Here is the whole method, published so you can check our work.

01 / BRIEF

One identical brief

We write one representative task per modality and publish the exact prompt, so every tool gets the same input.

02 / RUN

Every contender

That brief goes through every serious tool, on a current paid plan where needed. We capture the real output.

03 / SCORE

Fixed rubric

Each output is scored on the same five axes, so the numbers are comparable across the whole field.

04 / VERDICT

Best for X

We name the best tool for each kind of job, not a single ranking, because the right pick depends on the work.

The five-axis rubric

Every tool in a bake-off is scored on the same five axes. We weight them per modality (licensing matters more for commercial video than a personal sketch), and we publish the weighting on each bake-off.

AxisWhat it measures
Prompt adherenceHow faithfully the output matches the exact brief, without re-rolling or hand-holding.
Output qualityRaw fidelity and craft: detail, coherence, realism or style control, and freedom from obvious artifacts.
ControlHow much the tool lets you steer, edit, iterate, and lock a result, versus rolling the dice.
Price per outputThe real cost of a usable result, including credits, re-rolls, and plan tier, not just the headline price.
Licensing clarityWhether you can actually use the output commercially, and how clearly the tool tells you so.

Scores are our editorial judgment against the published brief. We do not accept payment to change one.

Freshness and re-runs

Creative-AI tools change pricing, limits, and model versions almost weekly. We date-stamp every bake-off with the month we tested, and we re-run it when a major model drops or a tool changes its pricing. A verdict that does not say when it was tested is a verdict you should not trust, here or anywhere.

What we will not do

We will not let a sponsor change a score, reorder a ranking, or remove a competitor. We will not claim an AI engine cites us until we can screenshot it. We will not fabricate a number we could not verify; if a price or limit is genuinely unverifiable, we say so on the page instead of guessing. The honesty floor is the point. A bake-off that can be bought is worthless to the people we serve.

Save
Dashboard