Free OCR API: free tiers and limits compared

Free OCR API tiers compared

Limits as of November 2026, taken from each provider's documentation and pricing pages. The free tiers count different things: Vision counts units (one per PDF page), Textract and Azure count pages, and OCR.space counts requests whatever the page count.

Which free OCR API to pick

Code: OCR a PDF with the FastOCR API

Any signed-in FastOCR account can create an API key and start right away. Install the Python SDK with pip install fastocr-sdk and set FASTOCR_API_KEY:

from fastocr_sdk import FastOCR

# Reads the key from the FASTOCR_API_KEY environment variable.
with FastOCR() as client:
    text = client.extract_text("scan.pdf")
    print(text)

Without the SDK, the same job is five HTTP calls: create, upload to a presigned URL, start, poll, download.

export FASTOCR_API_KEY="your_api_key"

# 1. Create a document
RESPONSE=$(curl -sS --fail-with-body -X POST https://api.fastocr.org/v1/documents \
  -H "Authorization: Bearer $FASTOCR_API_KEY" \
  -H "Idempotency-Key: $(uuidgen)" \
  -H "Content-Type: application/json" \
  -d '{"filename":"scan.pdf","size_bytes":'$(wc -c < scan.pdf | tr -d ' ')'}')
DOCUMENT_ID=$(echo "$RESPONSE" | jq -r .id)
UPLOAD_URL=$(echo "$RESPONSE" | jq -r .upload_url)

# 2. Upload the file to the presigned URL
curl -sS --fail-with-body -X PUT "$UPLOAD_URL" \
  -H "Content-Type: application/pdf" \
  --data-binary @scan.pdf

# 3. Start processing
curl -sS --fail-with-body -X POST https://api.fastocr.org/v1/documents/$DOCUMENT_ID/start \
  -H "Authorization: Bearer $FASTOCR_API_KEY"

# 4. Poll until the status is completed, partial, or failed
while true; do
  BODY=$(curl -sS --fail-with-body https://api.fastocr.org/v1/documents/$DOCUMENT_ID \
    -H "Authorization: Bearer $FASTOCR_API_KEY") || exit 1
  STATUS=$(echo "$BODY" | jq -r .status)
  case "$STATUS" in completed|partial|failed) break;; esac
  sleep 2
done
echo "Final status: $STATUS"

# 5. Get the download link, then download the text (a failed document has no output)
RESULT_URL=$(curl -sS --fail-with-body "https://api.fastocr.org/v1/documents/$DOCUMENT_ID/output?format=text" \
  -H "Authorization: Bearer $FASTOCR_API_KEY" | jq -r .url)
curl -sS --fail-with-body "$RESULT_URL" -o result.txt

Swap format=text for format=pdf to get a searchable PDF. The full reference is in the API docs.

Self-hosting Tesseract

Tesseract is the best-known open-source OCR engine (Apache 2.0). It has no per-page cost and keeps documents on your machines. The trade-off is setup, language packs, image clean-up and the servers to run it.

1. Install it: apt install tesseract-ocr or brew install tesseract.
2. Add languages: apt install tesseract-ocr-ara tesseract-ocr-hin.
3. Read an image to out.txt: tesseract scan.png out, or make a searchable PDF with tesseract scan.png out pdf.

Tesseract reads images, not PDFs, so scanned PDFs need converting to images first. See FastOCR vs Tesseract for how the two differ.