About

The bench

UncensoredLab scores uncensored AI image and video generators on a 0-100 scale. The bench tests 12 tools across seven weighted readings: realism at 22%, freedom at 18%, prompt control at 17%, consistency at 15%, value at 12%, video at 10%, and usability at 6%. Rankings are organized by job, instead-of pages, head-to-head comparisons, and detailed notes.

The seven readings

Each tool receives a scorecard built from 27 scored dimensions grouped into seven pillars. Realism measures output fidelity to physical and environmental accuracy. Freedom evaluates the range of permitted content and prompt types. Prompt control tests how precisely the generator follows instructions. Consistency checks output stability across seeds and sessions. Value weighs pricing against capability. Video assesses motion quality and temporal coherence. Usability covers interface speed, documentation, and workflow efficiency.

Every dimension carries an evidence note and source, displayed on each review page as a detailed scorecard. See the full breakdown at /method.

Evidence and sources

Scores draw from a shared evidence base: each platform's own documentation combined with the pages that rank on Google for the category. This material is distilled into the 27 scored dimensions. Every scorecard shows the evidence note and source behind each dimension. Scores are re-checked quarterly; any changes are logged on the page.

No platform pays for a score, a rank, or a listing. Links may carry referral tags, which never affect a score.

Who runs the bench

Founding editor Scott Dommett built the bench after two years inside the AI companion and generative adult-content category as a paying user of most products covered. Reach him via editor@uncensoredlab.com or the contact page.

Corrections

Scores are re-checked quarterly. When a score changes, the update is logged on the review page with the date and reason. If you spot an error in an evidence note or source, send details to editor@uncensoredlab.com.

Limits

This bench is adults only. Any platform permitting content involving minors or non-consensual imagery of real people is excluded and reported. The 12 tools listed represent the current scope; coverage is limited to generators that meet this threshold. The seven readings and their weights reflect what matters to working users of these tools, not a universal claim about all AI generation.