# TimmyOVO/deepseek-ocr.rs

Rust multi‑backend OCR/VLM engine (DeepSeek‑OCR-1/2, PaddleOCR‑VL, DotsOCR) with DSQ quantization and an OpenAI‑compatible server & CLI – run locally without Python.

Repository: https://github.com/TimmyOVO/deepseek-ocr.rs
Canonical: https://ross.abutalabs.com/products/deepseek-ocrrs
Language: Rust
License: Apache-2.0
License Family: permissive
Topics: candle, ocr, ocr-recognition, openai, rust
Last push: 2026-02-21T12:21:10+00:00

## Health v2 (maintenance only)
Score: 60/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 68, release rhythm 71, longevity 22
- inputs: {"age_days": 312, "days_push": 193, "days_rel": 193, "gap_med": 0.5, "n_releases_24m": 15}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2182, forks 169 (observed 2026-08-28T04:06:23.370117+00:00)

## What it is
A Rust implementation of the DeepSeek-OCR inference stack with multiple OCR/VLM backends (DeepSeek-OCR, PaddleOCR-VL, DotsOCR), DSQ quantization, a fast CLI, and an OpenAI-compatible HTTP server. It runs locally on CPU, Apple Metal, or CUDA without any Python runtime, built on Candle and Rocket.

## Use cases
- extract text from images and scanned documents locally
- run OCR without installing Python or conda
- serve an OpenAI-compatible OCR API endpoint
- convert PDFs and documents to markdown with layout awareness
- run vision-language models on Apple Silicon with Metal
- batch OCR on CPU-only machines with low memory
- embed OCR inference into native Rust applications

## When to choose
- you need local, Python-free OCR inference in Rust
- you want an OpenAI-compatible serving layer for OCR models
- you're on Apple Silicon or have a decent GPU and want high-fidelity document understanding
- you need a CLI for quick one-off OCR tasks

## When to avoid
- you need the broadest model ecosystem of Python/HuggingFace Transformers
- you only have limited RAM and need DotsOCR's heavy vision tower
- you need managed/cloud OCR services rather than self-hosted inference

## Facets
- artifact type: library
- maturity: active
- function: ocr, llm-inference, cli, http-server, machine-learning
- domain: computer-vision, image-processing, pdf, developer-tools, machine-learning
- platform: rust, windows, cli
- tags: deepseek-ocr, paddleocr-vl, dotsocr, candle, openai-compatible-api, vision-language-models, quantization, dsq, document-understanding, local-inference, natural-language-processing, macos, linux, gpu

## Member repositories
- TimmyOVO/deepseek-ocr.rs (main) score 60

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:23.370117+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 2026-08-30T02:48:16.301961+00:00, confidence not recorded.
  - readme: https://github.com/TimmyOVO/deepseek-ocr.rs (fetched 2026-08-28T04:06:23.370117+00:00, sha 08cc9f5e7f82)
- Data as of 2026-08-30T08:39:29.467469+00:00.