Loading...
Loading...
docTR (Document Text Recognition) is a deep learning-based OCR library by Mindee. It uses a two-stage approach: text detection followed by text recognition, both powered by neural networks. Particularly strong at detecting text in complex layouts with varying orientations.
OpenOCR routes requests to the best provider able to handle your file type and parameters, with fallbacks to maximize uptime.
Benchmark score
Latency
2.0s
Languages
15
Max Pages
50
License
Apache-2.0
Supported Formats
Strengths
Weaknesses
| Engine | Provider | Benchmark | Price |
|---|---|---|---|
| Datalab Convert (accurate) | Datalab | 77.2%✓ | 100 pages / $1$0.01/page |
| Meta: Llama 4 Maverick | OpenRouter | 71%✓ | ~1,560 pages / $1~$0.00064 per standard page |
| Anthropic: Claude Opus 4.5 | OpenRouter | 67.5%✓ | ~46.1 pages / $1~$0.0217 per standard page |
| Meta: Muse Spark 1.2 | OpenRouter | 65.5%✓ | ~103 pages / $1~$0.00975 per standard page |
| Mistral: Mistral Medium 3 | OpenRouter | 64.9%✓ | ~684 pages / $1~$0.00146 per standard page |
| OpenAI: GPT-5.6 Luna | OpenRouter | 64.8%✓ | ~816 pages / $1~$0.00123 per standard page |
| Mistral: Mistral Medium 3.1 | OpenRouter | 64.6%✓ | ~684 pages / $1~$0.00146 per standard page |
| Gemini 3 Flash Preview | Google Cloud | 63.6%✓ | ~184 pages / $1~$0.00543 per standard page |
| StepFun: Step 3.7 Flash | OpenRouter | 63.6%✓ | ~1,150 pages / $1~$0.000867 per standard page |
| Anthropic: Claude Opus 4.6 | OpenRouter | 63.4%✓ | ~46.1 pages / $1~$0.0217 per standard page |
| docTRcurrent | Open-OCR.com | 12.8%✓ | 500 pages / $1$0.002/page |
Top 10 by benchmark score, plus this engine. Compare all 174 engines
Capabilities