# AIPR — AI Peer Review

> AIPR gives reviewers and editors a calibrated AI first read of a research paper: scores across novelty, rigor, applicability, clarity, and citation quality, a citation audit, and an editable draft review. A human reviewer always stays in control of the final review, and submitted manuscript content is never used to train a model.

## Key pages
- [Home](https://aipr.pub/): What AIPR is and how to get a review.
- [Methodology](https://aipr.pub/methodology): How papers are scored, how citations are checked, and where humans stay in control.
- [Weekly rankings](https://aipr.pub/platform): AI-scored rankings of newly published preprints.
- [Preprint report](https://aipr.pub/trends): Weekly write-up of what the preprint scan turned up.
- [Preprint statistics](https://aipr.pub/stats): Preprint activity by field and country.
- [For institutions](https://aipr.pub/institutions): AI peer review pilots for journals and editorial teams.
- [Pricing](https://aipr.pub/pricing): Plans from guest passes to subscriptions.
- [Data policy](https://aipr.pub/privacy): No training on submitted content; retention windows and deployment options.

## Notes
- Reviews are AI-drafted and human-finalized; the model never sends output directly to an author.
- Scores run from 0 to 100 on each of five dimensions and combine into an overall score adjusted for model confidence.
- The weekly rankings compare papers within a cohort and are published only after a reviewer approves the review.