Private Beta

FastOCR API

Extract text from documents and structured data from bank statements. Upload a file, poll for completion, download the result.

Getting access

The API is in private beta. Access is allowlisted per account — you cannot generate a key until your account is approved.

1Contact us to request API access for your account.
2Once approved, create an API key in the developer console.
3Pass the key as Authorization: Bearer <key> on every request.

Document OCR

Upload a PDF, run OCR, and get extracted text or a searchable PDF back. Five requests from file to result.

quickstart-documents.sh

FASTOCR_KEY="your-api-key"
BASE="https://api.fastocr.org"

# 1. Create a document job
curl -s -X POST "$BASE/v1/documents" \
  -H "Authorization: Bearer $FASTOCR_KEY" \
  -H "Idempotency-Key: $(uuidgen)" \
  -H "Content-Type: application/json" \
  -d '{"filename":"scan.pdf","size_bytes":'$(wc -c < scan.pdf | tr -d ' ')'}' \
  -o doc.json

# 2. Upload the PDF to the presigned URL
curl -s -X PUT "$(jq -r '.upload_url' doc.json)" \
  -H "Content-Type: application/pdf" \
  -T scan.pdf

# 3. Start processing
DOC=$(jq -r '.id' doc.json)
curl -s -X POST "$BASE/v1/documents/$DOC/start" \
  -H "Authorization: Bearer $FASTOCR_KEY"

# 4. Poll until ready
while true; do
  STATUS=$(curl -s "$BASE/v1/documents/$DOC" \
    -H "Authorization: Bearer $FASTOCR_KEY" | jq -r '.status')
  case "$STATUS" in completed|partial|failed) break ;; esac
  sleep 3
done

# 5. Download the extracted text
URL=$(curl -s "$BASE/v1/documents/$DOC/output?format=text" \
  -H "Authorization: Bearer $FASTOCR_KEY" | jq -r '.url')
curl -o result.txt "$URL"

Output formats: ?format=text for plain text, ?format=pdf for a searchable PDF.

Idempotency: The Idempotency-Key header prevents duplicate jobs if a request is retried. Use a unique value per job.

Statuses: awaiting_upload → processing → completed or failed. Large documents may finish as partial with is_truncated: true.

Bank statement extraction

Upload a bank statement and extract transactions as structured data. Same five-step flow as documents. Accepts PDF, PNG, JPG, and WebP.

quickstart-bank-statements.sh

FASTOCR_KEY="your-api-key"
BASE="https://api.fastocr.org"

# 1. Create a bank statement job
curl -s -X POST "$BASE/v1/bank-statements" \
  -H "Authorization: Bearer $FASTOCR_KEY" \
  -H "Idempotency-Key: $(uuidgen)" \
  -H "Content-Type: application/json" \
  -d '{"filename":"statement.pdf","size_bytes":'$(wc -c < statement.pdf | tr -d ' ')'}' \
  -o stmt.json

# 2. Upload the file
curl -s -X PUT "$(jq -r '.upload_url' stmt.json)" \
  -H "Content-Type: application/pdf" \
  -T statement.pdf

# 3. Start extraction
STMT=$(jq -r '.id' stmt.json)
curl -s -X POST "$BASE/v1/bank-statements/$STMT/start" \
  -H "Authorization: Bearer $FASTOCR_KEY"

# 4. Poll until ready
while true; do
  STATUS=$(curl -s "$BASE/v1/bank-statements/$STMT" \
    -H "Authorization: Bearer $FASTOCR_KEY" | jq -r '.status')
  case "$STATUS" in completed|failed) break ;; esac
  sleep 3
done

# 5. Download as Excel
URL=$(curl -s "$BASE/v1/bank-statements/$STMT/output?format=xlsx" \
  -H "Authorization: Bearer $FASTOCR_KEY" | jq -r '.url')
curl -o transactions.xlsx "$URL"

Output formats: ?format=xlsx (Excel), ?format=csv, or ?format=json.

File limits: PDFs up to 100 MB. Images up to 20 MB.

Endpoints

Base URL: https://api.fastocr.org

Method	Path	Description
POST	/v1/documents	Create a document job
GET	/v1/documents	List documents
GET	/v1/documents/{id}	Check status
POST	/v1/documents/{id}/start	Start processing
GET	/v1/documents/{id}/output	Download result
DELETE	/v1/documents/{id}	Delete a document
POST	/v1/bank-statements	Create a statement job
GET	/v1/bank-statements	List statements
GET	/v1/bank-statements/{id}	Check status
POST	/v1/bank-statements/{id}/start	Start extraction
GET	/v1/bank-statements/{id}/output	Download result
DELETE	/v1/bank-statements/{id}	Delete a statement

List parameters: ?limit= (1–100, default 20), ?status=, ?cursor=, ?updated_since=.

List response: the array is named after the resource — { documents: [...], next_cursor } and { bank_statements: [...], next_cursor }. A next_cursor of null means there are no more pages.

Create body: filename (required), size_bytes (required), external_id (optional), sha256 (optional).

Errors

A failed job returns HTTP 200 with a terminal status and an error object. HTTP status describes the request, not the work.

error object

{
  "error": {
    "type": "processing_error",
    "code": "timeout",
    "message": "This file took too long to process. Please try again.",
    "retryable": true
  }
}

Match on code, never on message — messages change without notice.
New codes will be added. Treat any unrecognised code as processing_failed.
Only buy_pages reopens a failed job — buy credits, then call /start again on the same job.

Code	Type	When	Retryable
buy_pages	insufficient_credits	Out of page credits.	No — buy a pack, then call /start again on the same job.
unreadable_statement	invalid_request	Bank statement could not be read.	No
unreadable_document	invalid_request	PDF could not be read.	No
too_many_pages	invalid_request	File exceeds the page cap.	No
empty_file	invalid_request	File has no pages.	No
timeout	processing_error	Processing ran out of time.	Yes
processing_failed	processing_error	Anything else.	Yes

Ready to get started?

Request access, create a key, and run the quickstart above.

Request access Manage keys