Extracts text and structured JSON from German invoices and business documents.
Process German invoices, receipts, and forms locally or via EU cloud API.
German-OCR is a specialized document AI tool built on Qwen models to extract structured JSON and text from German business documents. It suits engineers, estimators, and project managers handling German subcontractor invoices, receipts, and technical forms.
It provides two specialized models for Vision-OCR (image to strict JSON) and text assistance tasks like translation and summarization. You can deploy it locally via Ollama and llama.cpp or connect to a managed EU-hosted cloud API.
It suits teams processing high volumes of German construction invoices and compliance paperwork who require local data privacy or cloud-based scaling. It does not suit workflows focused entirely on non-German languages or standard CAD document types.
Engineered for 100 percent valid JSON output on real German invoices.
Run locally via Ollama, llama.cpp, and Hugging Face, or use the managed cloud API.
Integrate via official Python, Node.js, and PHP packages.
Test form filling instantly using the client-side local browser demo.
The software is free under its open source license, while the hosted cloud API option requires paid credentials obtained via the provider's app portal.
880 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Yes, you can run the models locally using Ollama, llama.cpp, or Hugging Face backends without sending data to the cloud.
The tool supports Markdown, plain text, and strict JSON outputs tailored for data extraction.
See Zekai first in Google