MPDocBench-Parse - Heading Hierarchy: leaderboard
Recovering the heading hierarchy of multi-page documents as a Markdown heading tree; MPDocBench-Parse (Zhou et al., 2026). Tree-edit similarity from 0 to 100 against the annotated hierarchy; higher is better.
Metric: Tree-edit-distance similarity (0-100; predicted Markdown heading tree against the annotated heading hierarchy). Source: arxiv.org. Saturation forecast: Around 2032. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 VL 235B A22B | 42.34 | #477 |
| 2 | Qwen 3.6 Plus | 32.06 | #217 |
| 3 | Gemini 3.1 Pro (Preview) | 26.94 | #91 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=mpdocbench-parse-heading-hierarchy · How It Works · Data refreshed daily, snapshot 2026-11-12.