HighlightBench (Real-World) - Aggregation and Comparison: leaderboard
Metric: Exact-match accuracy (%) on the aggregation and comparison task family of the real-world subset (markup-rich tables from scientific papers with manually written questions); higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Flash | 72.2 | #237 |
| 2 | Claude Sonnet 4 | 60.7 | #194 |
| 3 | MiniCPM-V-2.6 | 17.5 | #825 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=highlightbench-real-world-aggregation-and-comparison · How It Works · Data refreshed daily, snapshot 2026-10-11.