MDPBench — leaderboard
Multilingual document parsing benchmark covering digital and photographed documents across diverse scripts, layouts, and languages.
Metric: Overall Score (%). Source: huggingface.co. Status: saturation imminent. 31 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 86.4 |
| 2 | Kimi K2.5 | 77.5 |
| 3 | Claude Sonnet 4.6 | 73.1 |
| 4 | Qwen 3 VL 8B Instruct | 68.3 |
| 5 | Qwen 3.5 9B | 65.7 |
| 6 | InternVL3.5-8B | 42.7 |
Interactive version: theaggregate.ai/benchmark?slug=mdpbench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.