How we build our rankings
Our rankings are built on independent public benchmarks and named public sources. Here is exactly how it works, and where the limits are.
Where the score comes from
No single person can exhaustively test every AI video tool — this space moves too fast and covers too many categories. We do not claim to have tested them all, and we do not publish numbers we cannot trace to a source. Every score on this site comes from one of the two routes below, and each review states which one applies.
Independent benchmarks
Where a public benchmark measures a model — Artificial Analysis, Arena.ai or TTS Arena — we take the published leaderboard figure and convert it to a 0-100 scale. These are blind comparisons run by third parties, not by us. Each review names the source and the date it was checked.
Named public sources
Many tools are not covered by any benchmark. For those we score relative to others in the same category, using named, dated public reviews and vendor documentation. These scores carry a Provisional or One source label so you can see how thin the evidence is.
The scoring criteria
Every tool page shows three numbers. Only the first one sets the ranking.
GraiLogic score
Output quality only. Where a benchmark covers a model, the score is its expected win rate against a typical model in the same category: 50 means it performs about as well as a typical model, and neither 0 nor 100 can occur. Scores from different categories are on different scales and are not comparable across categories.
Ease of use
Time from signup to first usable output, how much technical knowledge is required, and how clear the interface is. This is our own editorial judgement, shown for context. It does not change the score or the ranking.
Value for money
What you actually get at each price tier, how fast credits burn in practice, and the real cost per usable output after iteration. This is also our own editorial judgement, shown for context. It does not change the score or the ranking.
The community ranking
A second, separate ranking built entirely from scores submitted by real users is planned, with no editorial input. It is not live yet: there is currently no way to submit a rating on GraiLogic.
When it launches, it will go live for a tool only once that tool has received enough verified ratings to be statistically meaningful, and it will sit alongside our own ranking rather than replace it.
What gets a tool on the Blacklist
We maintain a Blacklist of tools we actively warn against. A tool can end up there for:
Misleading pricing
“Unlimited” plans that throttle or cap in practice. Low prices that require annual commitment buried in small print. Credits that expire without clear warning.
Predatory billing
Charging after free trials without clear notice. Refusing refunds for failed generations. Making cancellation deliberately difficult.
Output that doesn’t match claims
Marketing materials showing quality that cannot be reproduced on any paid plan. Demo videos that use footage not generated by the tool itself.
Active legal or safety risks
Tools facing litigation for copyright infringement, or tools whose outputs carry significant legal risk for creators using them commercially.
Blacklisting is not permanent. If a tool fixes the problem we flagged, we re-evaluate it and remove it from the list.
What we cover — and what we don’t
GraiLogic covers any tool with meaningful video creation, editing, or enhancement capabilities — whether video is its only function or one of many. If a tool can meaningfully help you make better video, it belongs here. What we don’t cover: tools where video is a minor or superficial feature that adds no real value to a video creator’s workflow.