How AI video rankings are scored
The ranking system separates model quality from access, usability, cost, open-source health, web authority, and source confidence.
Default Score Blend
The Overall Consumer Pick Score weights creative quality most heavily, then blends usability, control, speed and cost, access, commercial safety, popularity, and open ecosystem data. Popularity and SEO signals are secondary: they help explain adoption, but they do not replace quality evidence.
Provisional Scores
Some current public scores are provisional seed estimates until the fine-grained scoring task finishes filling the scoring JSON. Provisional values stay labeled in generated pages and are meant to keep the site useful while deeper benchmark, prompt-suite, and source-confidence work continues.
Scoring Dimensions
Each family normalizes to a 0-100 scale before the public blend is calculated.
| Dimension | What It Covers |
|---|---|
| Creative Quality | Prompt adherence, temporal coherence, motion quality, visual quality, controllability, and production readiness. |
| Consumer Usability | Time to first video, UI clarity, iteration workflow, export, mobile fit, learning support, billing transparency, and localization. |
| Control | Reference support, camera control, character consistency, keyframes, multi-shot/storyboard tools, style controls, and editing depth. |
| Speed And Cost | Generation latency, queue behavior, free tier, subscription value, API pricing, and duration/resolution tradeoffs. |
| Access And Availability | Web, mobile, API, region availability, waitlist status, uptime, and current product reliability. |
| Trust And Commercial Safety | Commercial rights, watermark policy, privacy posture, moderation, enterprise maturity, and source confidence. |
| Popularity Signal | Search demand, review evidence, web visibility, third-party listings, and developer or consumer traction. |
| Open Ecosystem | Open/closed marking, GitHub stars, repository freshness, license clarity, Hugging Face or local inference support, and community tooling. |
Data Sources
Inputs come from official model pages, product documentation, release notes, API docs, company pages, benchmark references, manual product testing notes, comparison datasets, and curated news/version monitoring. External claims remain linked back to public source URLs where available.
Open Or Closed
Every model is marked by access type such as closed web, closed API, aggregator, open weights, research only, or local/self-hosted. Open models receive additional open ecosystem scoring so a builder-facing option is not judged only like a consumer web app.
GitHub And Open Metrics
For open models, repository star counts, activity freshness, license clarity, release cadence, forks, ecosystem tooling, and runnable inference paths inform the Open Ecosystem Score. These metrics are directional signals, not a direct creative-quality score.
SEO And Web Authority
Search visibility, crawlable site size, robots and sitemap availability, authoritative source coverage, and web authority records help estimate product discoverability and traction. They are capped as popularity or evidence signals so a famous brand does not automatically outrank a better generator.
News And Version Monitoring
Release posts, documentation changes, company milestones, model-version events, and sourced news links update freshness, confidence, access status, and product context. They can change a score when they confirm a new model, new API route, pricing shift, or meaningful capability change.
Evidence Confidence
Confirmed first-party evidence carries more weight than stale, inferred, or third-party-only references. Missing evidence reduces confidence before it changes the headline score, making incomplete pages honest without hiding useful candidates.
What To Read Next
The rankings and comparison pages use this scoring policy in crawlable HTML.