Two adversarial reviewers
gpt-6.1-sol and gemini-4-argon each read the whole paper as a peer reviewer who looks for what is wrong with it.
XReview
Have a paper reviewed by models from three AI companies, OpenAI, Google and Anthropic. You get a score from 0 to 100 and a report with findings, evidence and disagreements between models.
Sign in to review a paperHow a paper is reviewed
Checked against the paper
Every objection is held against what the paper itself says, and its severity is judged.
One report
Claude Sonnet 5.5 combines the reviews into one editorial judgment. The report gives a score, the findings and disagreements between models.
Reading the score
The score runs from 0 to 100 in five bands, said in words.
- 0–20incoherent or self-contradictory
- 21–40major rigor gaps, central claims unsupported by the paper’s own logic
- 41–60mixed: real content, but one named defect keeps it from holding together
- 61–80sound for its kind, with addressable points
- 81–100internally rigorous for its kind; remaining points are polish
Reviewed papers
Papers with public text and a completed review, ranked by the score of the newest review.
| Heaviside's Rejected Wave: The Compressional Electric Wave of 1897 Read Against Modern Electrodynamics | Wilhelm | 72/100 | 30 Oct 2026 |
|---|---|---|---|
| The Deleted Degrees of Freedom: A Case for Potential-Primary Electrodynamics | Wilhelm | 24/100 | 28 Oct 2026 |