# DeepSeek OCR-2: vision-language OCR model on Hugging Face

> DeepSeek published DeepSeek-OCR-2, a dedicated image-text-to-text OCR model on Hugging Face under the Apache-2.0 license. The repository uses custom transformer code and carries tags for multilingual support, vision-language, and feature extraction. Two associated arXiv papers are referenced.

| | |
|---|---|
| **Tool** | DeepSeek |
| **Version** | — |
| **Kind** | release |
| **Published** | 2026-01-27 |
| **Observed** | 2026-08-11 |
| **Significance** | 1/5 |
| **Breaking** | no |
| **Categories** | capability |

> **Note:** this classification is below our confidence threshold and is pending human review.

## What changed


- Pipeline: image-text-to-text (vision-language OCR)
- Library: transformers with custom_code
- Tags: multilingual, vision-language, ocr, feature-extraction, deepseek_vl_v2
- License: Apache-2.0
- Referenced papers: arxiv:2601.20552, arxiv:2510.18234
- Repository created 2026-01-27, last modified 2026-02-03


## Sources

- [hf_model](https://huggingface.co/deepseek-ai/DeepSeek-OCR-2) — retrieved 2026-08-11


## Community

_No curated reactions recorded. Facts and community takes are kept in separate layers
and never blended._

---
Canonical: https://changelogs.info/deepseek/deepseek-ocr-2-vision-language-ocr-model-on-hugging-face
Entity: https://changelogs.info/deepseek
Event ID: `evt_2026-01-27_deepseek_deepseek-ai-deepseek-ocr-2`
Licence: event synthesis © changelogs.info, CC BY 4.0. Linked sources belong to their vendors.