# deepseek-ai/DeepSeek-OCR-2

Visual Causal Flow

Repository: https://github.com/deepseek-ai/DeepSeek-OCR-2
Canonical: https://ross.abutalabs.com/products/deepseek-ocr-2
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2026-02-03T00:34:18+00:00

## Health v2 (maintenance only)
Score: 44/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 65, release rhythm 35, longevity 15
- inputs: {"age_days": 218, "days_push": 212, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 3379, forks 304 (observed 2026-08-28T04:07:58.557798+00:00)

## What it is
DeepSeek-OCR 2 is an open-source vision-language model and inference toolkit implementing 'Visual Causal Flow' for optical character recognition and document understanding. It provides scripts for running inference on images and PDFs via vLLM or Hugging Face Transformers.

## Use cases
- extract text from images with ocr
- convert pdf documents to text
- parse scanned documents
- run ocr model locally on gpu
- benchmark document parsing on OmniDocBench
- batch process images for text extraction

## When to choose
- you need high-quality OCR or document-to-text conversion with a modern vision-language model
- you want GPU-accelerated batch or concurrent PDF OCR via vLLM
- you're researching visual encoding approaches like Visual Causal Flow

## When to avoid
- you need lightweight CPU-only OCR without a GPU
- you need a simple hosted OCR API rather than self-hosted model inference
- you lack the CUDA/torch environment the project requires

## Facets
- artifact type: library
- maturity: active
- function: ocr, machine-learning, llm-inference, pdf
- domain: computer-vision, image-processing, pdf, deep-learning, artificial-intelligence
- platform: python
- tags: vision-language-model, document-parsing, vllm, transformers, visual-causal-flow, deepseek, gpu, linux, docker

## Member repositories
- deepseek-ai/DeepSeek-OCR-2 (main) score 44

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:58.557798+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:39:41.811883+00:00, confidence not recorded.
  - readme: https://github.com/deepseek-ai/DeepSeek-OCR-2 (fetched 2026-08-28T04:07:58.557798+00:00, sha 5032c9384548)
- Data as of 2026-08-30T08:39:29.467469+00:00.