CoCoReviewBench - Clarity: leaderboard

CoCoReviewBench (2026) evaluates AI peer reviewers on ICLR and NeurIPS papers against category-specific human review opinions that were filtered for correctness using reviewer, author and meta-review discussions. This board measures how clear and well structured the comments are on a 1 to 5 judge scale, shown as the difference from the leave-one-out human reviewers, so positive values beat the humans; higher is better.

Metric: Category-level clarity score (coherence, readability and structure of the comments) on the 1-5 judge scale, printed as the difference from the human reference reviews (4.15, leave-one-out: one human review scored against the others), so a positive value is better than the human reviewers and the range is -3.15 to +0.85; AI reviews of 1,300 ICLR and NeurIPS papers sampled from CoCoReviewBench (100 per year, original score distribution kept), each split into atomic opinions and compared category by category with the filtered human opinions by a GPT-5-Mini judge at temperature 0, general models prompted with the AI Scientist review prompt at temperature 0.6; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 17 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1CycleReviewer-70B0.48#1880
2Gemini 3 Pro0.42#145
3Gemini 3 Flash0.41#175
4OpenReviewer-8B0.37
5nemotron-3-nano-30B-a3BReasoning0.36#1946
6Llama 3.3 70B0.33#1514
7GPT-5.20.32#185
8Qwen 3 32BThinking0.32#972
9GPT-5 Mini0.31#364
10CycleReviewer-8B0.3
11QwQ-32B0.27#907
12Qwen 2.5 7B0.17#2929
13DeepReviewer-14B0.17#844
14Llama 3.1 8B0.08#5196
15Qwen 3 8BNon-reasoning-0.07#1873

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=cocoreviewbench-clarity · How It Works · Data refreshed daily, snapshot 2026-11-21.