olmOCR Bench: leaderboard
OCR benchmark from Allen AI evaluating document understanding and text extraction accuracy across document types.
Metric: Overall Score. Source: github.com. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Chandra OCR 0.1.0* | 83.1 | |
| 2 | Infinity-Parser 7B* | 82.5 | |
| 3 | olmOCR v0.4.0 | 82.4 | |
| 4 | PaddleOCR-VL* | 80 | |
| 5 | Marker 1.10.1 | 76.1 | |
| 6 | DeepSeek-OCR | 75.7 | |
| 7 | MinerU 2.5.4* | 75.2 | |
| 8 | Mistral OCR API | 72 | |
| 9 | Nanonets-OCR2-3B | 69.5 |
Interactive version: theaggregate.ai/benchmark?slug=olmocr-bench · How It Works · Data refreshed daily, snapshot 2026-10-20.