SciZoom - TLDR Generation: leaderboard
Metric: ROUGE-L (0-100) of a one-sentence TL;DR (64 new tokens), scored against the author-provided OpenReview TL;DR, zero-shot from the full text of ML/NLP papers (NeurIPS, ICLR, ICML, EMNLP 2020-2025 on OpenReview, front-back truncated to the context window), greedy decoding; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 7 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3 8B Instruct | 28.8 | #1115 |
| 2 | Llama 3.1 8B Instruct | 28.6 | #1018 |
| 3 | DeepSeek R1 | 24 | #245 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=scizoom-tldr-generation · How It Works · Data refreshed daily, snapshot 2026-10-11.