Guides
Processing modes
Trade off accuracy against speed and cost.
Each extraction runs in one of two modes, set with the processingMode
parameter.
| Mode | Value | Best for |
|---|---|---|
| Highest Quality (default) | highest_quality | Complex, dense, or business-critical documents. |
| Fastest | fastest | High-volume, simpler documents where speed and cost matter most. |
curl https://app.docr.dev/api/v1/extract \
-H "Authorization: Bearer $DOCR_API_KEY" \
-F "file=@receipt.jpg" \
-F "processingMode=fastest"The model actually used is returned as modelUsed on the extraction, so you can
audit which tier handled each document.
Images and scanned PDFs always use a vision-capable model regardless of mode, since they require OCR.