SIN-Bench - SIN-Summary: leaderboard
Evidence-anchored summarization in SIN-Bench (Ren et al., 2026): the model summarizes a whole scientific paper given as interleaved text and figures, citing a figure and text anchor for each claim. The score averages evidence matching judged by Qwen3-8B, relevance F1 and Kendall-tau order similarity against the reference anchors, in percent; higher is better.
Metric: Evidence-Anchored Summary Score (%; mean of evidence matching, relevance F1 and Kendall-tau logic). Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 61 | #163 |
| 2 | Gemini 3 Pro (Preview) | 60 | #111 |
| 3 | Gemini 2.5 Pro | 59.3 | #262 |
| 4 | Claude Sonnet 4.5 (Thinking) | 55.2 | #244 (Claude Sonnet 4.5) |
| 5 | Grok 4 | 45.8 | #306 |
| 6 | Qwen 3 VL 8B (Thinking) | 43.9 | #887 |
| 7 | Qwen 3 VL 30B A3B (Thinking) | 43.2 | #710 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=sin-bench-sin-summary · How It Works · Data refreshed daily, snapshot 2026-11-12.