PaperMind - Critical Assessment: leaderboard
PaperMind (2026) tests reasoning over real scientific papers from seven domains, with answers rated 1 to 5 by a GPT-4o judge against references taken from the papers. Here the model, as a tool-using agent, answers substantive questions that OpenReview reviewers raised about computer-science submissions, in rebuttal style; higher is better.
Metric: Judge score for answering, rebuttal-style, substantive reviewer questions on real OpenReview computer-science submissions (questions that drew author responses and score increases), as a smolagents ReAct agent with tools; rated by a GPT-4o judge on a 5-point scale (1-5) against the reviewer questions; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 7 models tracked. Updated 14 Nov 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 2.64 | #276 |
| 2 | GPT-4o Mini | 2.57 | #1667 |
| 3 | Claude 3 Haiku | 2.51 | #2663 |
| 4 | Claude 3.5 Sonnet | 2.43 | #752 |
| 5 | Qwen 3 VL 4B Instruct | 2.42 | #1332 |
| 6 | Gemma 3 4B (IT) | 1.95 | #4139 |
| 7 | Phi-3.5-Vision Instruct | 1.26 | #6268 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=papermind-critical-assessment · How It Works · Data refreshed daily, snapshot 2026-11-21.