D
dots.ocr
Description
dots.ocr is an open multilingual document parser with 3 billion parameters. One model finds the layout of a page and reads the content: text, tables (output as HTML), formulas (output as LaTeX) and the reading order. It can also read a given area of the page. The authors' own benchmark covers 100 languages. The weights are under the MIT license and are not gated, and a Docker image is provided. Hardware requirements are not stated; the examples use CUDA. The authors note limits: complex tables and formulas are not perfect, pictures inside documents are not parsed, large PDF batches are not optimized, and some repeated special characters can cause endless output.
Pros and cons
Pros
- MIT license, commercial use allowed
- Layout, text, tables and formulas in one model
- Tables as HTML and formulas as LaTeX
- Docker image provided
Cons
- Hardware requirements are not stated, examples use CUDA
- Complex tables and formulas are not perfect, per the authors
- Not optimized for large PDF batches
- Handwriting is not listed among its features
Key facts
- License
- MIT
- Commercial use
- Allowed
- Developer
- rednote-hilab (dots studio)
- Checked on
- Oct 20, 2026
Editorial score
How strongly the AI Expertise editors recommend this open model for each of its tasks right now, on a 0–10 scale. The score combines output quality, capabilities, ease of launch, license openness and freshness.