LitReviewBench - Claim Support: leaderboard
LitReviewBench (Zhao et al., 2026) ranks literature review systems by about 3,000 blind pairwise judgments of matched domain experts on survey drafts for AI research topics, with human-written drafts in the pool. This board rates how reliably key claims are backed by relevant citations. Elo-scale rating from Bradley-Terry aggregation; higher is better.
Metric: Expert-preference rating (Elo scale, Bradley-Terry aggregation of about 3,000 anonymized pairwise battles judged by matched domain experts on D2 claim support by relevant citations; literature review drafts on AI topics; the human-draft baseline is in the rating pool). Source: arxiv.org. Saturation forecast: Around January 2027. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.5 | 1032.8 | #152 |
| 2 | Grok 4 | 908.5 | #306 |
| 3 | Qwen 3 235B A22B | 826.3 | #622 |
| 4 | Gemini 2.5 Pro | 601.8 | #262 |
| 5 | GLM-4.6 | 553.6 | #485 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=litreviewbench-claim-support · How It Works · Data refreshed daily, snapshot 2026-11-12.