NovBench (RAG) - Correctness: leaderboard
NovBench novelty assessment of EMNLP 2023 papers with a retrieval-augmented prompt with the five most similar ACL, EMNLP and NAACL 2019-2022 titles and abstracts. Correctness is the agreement between the positive, neutral and negative balance of the generated statements and that of the human reviewers, from 0 to 1, higher is better.
Metric: Correctness (DistAcc, 0-1): one minus half the L1 distance between the positive, neutral and negative proportions of the generated statements and of the human comments; retrieval-augmented prompt with the five most similar ACL, EMNLP and NAACL 2019-2022 titles and abstracts; 1,684 EMNLP 2023 papers: from the novelty-description sentences of the paper introduction (selected by GPT-5) the model writes a novelty evaluation as positive, neutral and negative statements, compared with the expert reviewers novelty comments (extracted by GPT-4o-mini, then deduplicated, consolidated and sorted by sentiment by GPT-4o); greedy decoding, at most 4,096 tokens; higher is better. Source: arxiv.org. Saturation forecast: Not forecast. 18 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | SEA-S | 0.674 | |
| 2 | Qwen 3 8B | 0.6737 | #1873 |
| 3 | Qwen 3 32B | 0.6604 | #972 |
| 4 | GPT-5 | 0.6453 | #178 |
| 5 | GPT-OSS-20B | 0.643 | #1183 |
| 6 | DeepSeek R1 Distill Qwen 14B | 0.6417 | #2771 |
| 7 | DeepSeek R1 Distill Llama 70B | 0.626 | #1669 |
| 8 | Qwen 3 14B | 0.6245 | #1269 |
| 9 | GPT-OSS-120B | 0.6027 | #704 |
| 10 | SEA-E | 0.5965 | |
| 11 | Gemini 2.5 Flash | 0.5923 | #463 |
| 12 | DeepReviewer-14B | 0.5246 | #844 |
| 13 | DeepSeek-R1-8B (NovBench checkpoint unspecified) | 0.522 | |
| 14 | CycleReviewer-70B | 0.4892 | #1880 |
| 15 | DeepReviewer-7B | 0.3907 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=novbench-rag-correctness · How It Works · Data refreshed daily, snapshot 2026-11-21.