ResearchQA (Open Paper) - Refusal Correctness: leaderboard
Metric: Refusal correctness (%; share of the adversarial false-premise questions in the sample answered with a refusal or with only grounded citations, citation-grounded answers from the Open Paper chat-with-paper harness to a 100-row evenly spaced sample of 6,211 single-paper questions over 494 open-access papers, citations checked by a normalizing substring matcher against the extracted paper text; higher is better). Source: arxiv.org. Saturation forecast: Around December 2026. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 87 |
| 2 | GPT-4.1 | 87 |
| 3 | GPT-5.4 | 78.3 |
| 4 | GLM-4.7 | 78.3 |
| 5 | Claude Opus 4.7 | 73.9 |
| 6 | Claude Haiku 4.5 | 73.9 |
| 7 | Gemini 3 Flash (Preview) | 65.2 |
| 8 | GPT-OSS-120B | 47.8 |
Interactive version: theaggregate.ai/benchmark?slug=researchqa-open-paper-refusal-correctness · How It Works · Data refreshed daily, snapshot 2026-09-29.