ICLR 2027

Every ICLR 2027 submission, reviewed ten times by AI

We reviewed every public ICLR 2027 submission ten independent times with the same fixed prompt. This page shows the score distribution across submissions.

All public submissions have now been reviewed. To avoid influencing the ongoing review process, we are publishing only the score distribution for now: the overall mean, the distribution of scores, differences between tracks, and the distribution within each track. Per-paper results will be released after the ICLR 2027 review period ends.

See the full score distribution →Per-paper results: after the review period
AI scores cannot replace human evaluation. On our ICLR 2025/2026 sample, the strongest AI model we tested predicts acceptance with an AUC of about 0.72; for comparison, page count alone reaches 0.62. This site only shows how AI reviewing sees the field: if every reviewer were purely AI, would my paper get in? And where does the 6 I got before submitting actually stand?

This page is only for looking at the distribution. The reviews here were generated with GPT-6-Luna, a smaller model, so that every submission could be reviewed ten times. Reviews you order on HackMyPaper run on GPT-6-Sol with medium reasoning, close to how most people try an AI review of their own paper, and can be repeated up to 10 times (Full review) to show how much the score moves. Every new account gets 5 free credits, enough for one Standard review.

Try it on your paper →
Paper IDs covered1–10,0006,133 papers reviewed
Average score (mean of 10)5.47median 5.46
Cut-off at 25% acceptance6.21top 25% of papers score at or above this
Papers at or above 631%within-paper SD 0.25

Distribution of paper scores

Each paper's mean of its ten reviews, 0.5-point bins

123456789

The highlighted bin contains the 25%-acceptance cut-off (6.21). Scores are on the ICLR 1–10 scale; the model is asked for a non-integer score, so paper means are rarely whole numbers.

Reviews are simulated by an AI model and are not ICLR reviews. They carry no information from the official review process and are provided for research on automated review. Want the same review on your own draft? Upload it on HackMyPaper.