OpenExtract

Self-hosted, drop-in replacement for AWS Textract. Change one line — keep your data.

client = boto3.client("textract",
    endpoint_url="http://localhost:8080",  # the only change
    region_name="us-east-1", aws_access_key_id="x", aws_secret_access_key="x")
client.detect_document_text(Document={"Bytes": img})   # same boto3 code
What does Textract cost you — and what would you save?
–
AWS Textract / mo
–
OpenExtract (local) / mo*
–
You keep / year
Get it on GitHub →
Get notified when the hosted API + Pro features launch

We'll email you when the managed endpoint or paid Pro license is ready. Two short questions help us prioritize which one to build first.

Thanks — we'll be in touch.

No form, no problem — email us directly.

* Local cost ≈ $0.09 / 1,000 pages amortized on a single GPU; CPU/Tesseract baseline is free. Apache-2.0 · no telemetry · data never leaves your machine. Pricing from published mid-2026 AWS rates; your mileage varies — run the benchmark on your own pages.