MPDocBench-Parse - Figure Extraction: leaderboard
Metric: Figure F1 (0-100; figure boxes matched by IoU, averaged over documents). Source: arxiv.org. Saturation forecast: Around August 2027. 17 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.6 Plus | 64.8 |
| 2 | Qwen 3 VL 235B A22B | 63.25 |
| 3 | Gemini 3.1 Pro (Preview) | 59.08 |
Interactive version: theaggregate.ai/benchmark?slug=mpdocbench-parse-figure-extraction · How It Works · Data refreshed daily, snapshot 2026-09-26.