Language Support

Using Languages

Pass settings.language with your OCR request as a standard 2-letter ISO 639-1 code (e.g., en, fr, zh). The router handles code conversion automatically. Engine-native multi-language hints can also be passed via the top-level lang array (e.g. ["eng", "fra"]).

# OCR with language hint
curl -X POST https://api.open-ocr.com/v1/ocr \
  -H "Authorization: Bearer sk-ocr-your-api-key" \
  -H "Idempotency-Key: $(uuidgen)" \
  -H "Content-Type: application/json" \
  -d '{
    "engine": "local-tesseract",
    "settings": { "language": "fr" },
    "input": {
      "type": "url",
      "url": "https://tesseract-ocr.github.io/tessdoc/images/eurotext.png"
    }
  }'

Note: If no language is specified, the router defaults to English (en). OpenOCR-hosted engines such as Tesseract do not detect language, so non-English documents must include the language code or they will return garbled output for non-Latin scripts. Some engines, including Mistral OCR, ignore language hints and detect the language automatically.

Automatic Code Mapping

Different engines use different language code formats. OpenOCR normalizes them:

Language Support Matrix

Top 20 languages and their engine support. All engines support more languages than listed — this covers the most commonly requested ones.