Skip to content
  • Pricing
New Transkribus API · Launching June 2026

One API for any text. Printed or handwritten.

The Transkribus API recognises printed text, handwriting, and everything in between — in 100+ languages. But it's more than an API endpoint. Behind it sits a complete ecosystem: a visual editor for training data, no-code model training, powerful out-of-the-box models, and 300+ community models for the long tail.

EU-hosted (Austria) · GDPR-compliant · Async & long polling · Enterprise plans available

recognize.py
import requests

resp = requests.post(
    "https://transkribus.eu/processing/v2/processes",
    headers={"Authorization": f"Bearer {TOKEN}"},
    json={
        "config": {"modelId": 38230},
        "image": {
            "imageUrl": "https://your-archive.org/scan-001.jpg"
        }
    }
)
print(resp.json()["processId"])

The ecosystem

Not just an endpoint — a complete text recognition platform

Most APIs give you a black box. Transkribus gives you the tools to make the box work for your documents.

Upload

Send documents via API or web editor

Recognise

Out-of-the-box or custom models

Train

Visual editor + no-code ML training

Integrate

Async, long polling, or batch

Export & Publish

PAGE XML, ALTO, TEI, plain text — or publish directly as a searchable site

What makes this different

An API backed by a full platform — not a standalone endpoint.

Out-of-the-box models

The Super Models handle the most common scenarios without any setup. Modern print, historical print, and general handwriting across major European languages — just pick a model ID and go. For less common scripts and languages, browse 300+ community-trained models covering everything from medieval Latin to Ottoman Turkish.

Super Models for high-accuracy general recognition
300+ community models for specific scripts, centuries, and languages
Switch models per request — no configuration needed
models.sh
# Use a Super Model for general handwriting
curl -X POST .../processes \
  -d '{"config": {"modelId": 38230}, ...}'

# Use a community model for Ottoman Turkish
curl -X POST .../processes \
  -d '{"config": {"modelId": 47291}, ...}'

# Use your own custom-trained model
curl -X POST .../processes \
  -d '{"config": {"modelId": 98012}, ...}'

OCR + HTR

Printed text, handwriting, or both on the same page

Standard OCR engines handle printed text well but fail on handwriting. Transkribus was built for both from the start. The same API call processes a typed invoice, a handwritten letter, or a mixed page with print and marginalia. You don't need separate pipelines for different document types.
Modern and historical print — including Fraktur and blackletter
Handwriting from any century — cursive, block, connected scripts
Mixed pages — print, handwriting, stamps, annotations in one pass
Layout analysis handles columns, tables, marginalia, and complex pages
Handwriting and print recognition

Case study

FromThePage: AI-assisted crowdsourcing via API

FromThePage integrated the Transkribus API to add an AI Assist feature to their crowdsourcing platform. Volunteers see AI-generated transcriptions overlaid on document images — faster to transcribe, humans stay in the loop. Used by Harvard, Stanford, the British Library, and dozens of heritage institutions.
API integration completed in weeks
AI pre-transcription speeds up volunteer work
Used by major research institutions worldwide
FromThePage AI Assist powered by Transkribus API

Try it yourself

Upload a document — printed or handwritten — and see the API response in real time.

Drag an image here

Select a file...

PNG or JPEG, max 10 MB

Documentation

Everything you need to integrate the Transkribus API.

Authentication

The Transkribus API uses the OpenID Connect protocol for authentication. Obtain an access token, then include it as a Bearer token in all API requests.

OpenID Connect configuration: https://account.readcoop.eu/auth/realms/readcoop/.well-known/openid-configuration

EndpointsOpen Swagger UI →

Good to know

API credits cost 50% less than Transkribus UI credits. Pay-per-use billing is available upon request.

To use a private model, your API account must be added to the collection where the model is linked.

Accepted formats: JPEG, TIFF, PNG. Maximum file size: 20 MB per image.

200M+ pages transcribedREST APIGDPR compliantEU sovereign infra

Built on trust, hosted in Europe.

Your data stays on our infrastructure in Austria — no US cloud, no Big Tech dependencies.

EU-hosted & GDPR-compliant

All data processed and stored in Austria. Full compliance with European data protection regulations.

Enterprise-ready

Production infrastructure handling millions of pages. On-premises deployment coming soon for full data sovereignty. Dedicated support and SLA options available.

Cooperative, not corporate

READ-COOP is a European cooperative owned by 250+ member institutions. No vendor lock-in, no exit strategy.

Start building with the Transkribus API

Get API access with a free account. For enterprise workloads and volume pricing, talk to our team.

REST APIenterprise-ready
300+AI models
EU-hostedGDPR-compliant