MMDocBench: leaderboard
Fine-grained multimodal document understanding benchmark with OCR-free VQA, grounding, and document reasoning tasks.
Metric: Exact Match All (%). Source: mmdocbench.github.io. Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 71.99 | #343 |
| 2 | Gemini 1.5 Pro | 71.61 | #445 |
| 3 | Claude 3.5 Sonnet | 69.25 | #350 |
| 4 | Qwen 2.5 VL 7B Instruct | 65.79 | #654 |
| 5 | MiniCPM-V-2.6 | 51.32 | #836 |
| 6 | Qwen 2 VL 7B Instruct | 45.77 | #826 |
| 7 | InternVL2-8B | 43.22 | #817 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mmdocbench · How It Works · Data refreshed daily, snapshot 2026-10-20.