HyperAPI is a set of document intelligence APIs for invoices, receipts, bank statements, contracts and other finance documents. You send a PDF or image and get back clean text, typed fields, a document type or redacted pages. It works from a backend service, a dashboard or an AI agent.
Each endpoint takes a document_key from a single upload, so you can chain them without uploading the file again.
POST /v1/parseOCR to clean text and markdown with page geometry, line and word boxes, layout regions and table parsing. advanced mode adds per-page structured HTML and markdown.
POST /v1/extractTyped entities and line items from invoices and receipts. For any other document, pass your own JSON schema.
POST /v1/classifyReturns a document type with a confidence score and ranked candidates. You can add your own classes and descriptions.
POST /v1/splitBreaks a bundled PDF into separate logical documents, with a page range, label and confidence for each.
POST /v1/redactCovers PII with black boxes or replaces it with synthetic values. You can add your own entity types.
Python 3.9+. The official SDK installs from GitHub.
hk_live_ are for production and keys that start with hk_test_ are for testing.HYPERAPI_KEY.# install (the package installs as hyperapi-sdk and imports as hyperapi) pip install "hyperapi-sdk @ git+https://github.com/hyprbots/hyperapi-sdk.git" export HYPERAPI_KEY=hk_test_...
from hyperapi import HyperAPIClient with HyperAPIClient() as client: # reads HYPERAPI_KEY doc = client.process("invoice.pdf") # parse + extract in one call print(doc["result"]) text = client.parse("invoice.pdf") # OCR text + layout kind = client.classify("bundle.pdf") # document type + confidence parts = client.split("bundle.pdf") # page ranges per document safe = client.redact("statement.pdf", mode="deidentify")
pip install hyperapi. That name on PyPI belongs to an unrelated project. Use the GitHub install above.For FastAPI, aiohttp or agent loops, use AsyncHyperAPIClient and fan calls out with asyncio.gather. Every endpoint also has a submit_* variant that returns a job for bulk or long-running work. Batches are available through create_batch_from_files.
The flow has three steps: request an upload URL, PUT the bytes, then call any endpoint with the document_key. The key is valid for 24 hours.
# 1. request a presigned upload URL curl -X POST https://apis.hyperbots.com/v1/documents/upload \ -H "X-API-Key: $HYPERAPI_KEY" -H "Content-Type: application/json" \ -d '{"filename":"invoice.pdf","content_type":"application/pdf","content_length":184213}' # -> {"document_key":"550e8400-...","upload_url":"https://...","expires_in":900} # 2. upload the file to upload_url with an HTTP PUT (no API key needed) # 3. extract structured fields curl -X POST https://apis.hyperbots.com/v1/extract \ -H "X-API-Key: $HYPERAPI_KEY" \ -F "document_key=550e8400-e29b-41d4-a716-446655440000"
{
"status": "success",
"task": "extract",
"model_used": "HyperLM",
"result": {
"entities": {
"invoice_number": "INV-2024-0042",
"date": "2024-01-15",
"due_date": "2024-02-15",
"vendor_name": "Acme Supplies Ltd",
"total_amount": "1,250.00",
"currency": "USD"
},
"line_items": [ ... ]
}
}
# any document, your schema curl -X POST \ "https://apis.hyperbots.com/v1/extract?category=non_financial" \ -H "X-API-Key: $HYPERAPI_KEY" \ -F "document_key=..." \ -F 'schema={"patient_name":null,"active":false}' # dense tables and forms # add ?parse_mode=advanced (Plus, Pro, Custom) # process asynchronously and get a job id -H "X-Async: true"
Start free. Paid plans charge overage to prepaid credits at the per-page rates shown above. Results are cached per document for 24 hours, and cached results are not billed.
| Plan | Price | Pages / month | Rate limit | Highlights |
|---|---|---|---|---|
| Free | $0 | 30 | 1 req/min | All core APIs, fast parse, no credit card |
| Starter | $1 | 300 | 10 req/min | Core APIs, email support |
| Plus | $25 | 10,000 | 30 req/min | Advanced parse, team workspace with 3 seats |
| Pro | $99 | 50,000 | 60 req/min | Priority routing, up to 10 seats, priority support |
| Custom | Talk to us | Volume | Dedicated | BYOK, IP allowlist, human-in-the-loop review |
Current prices and promotions are on the pricing page.
GET /v1/limits returns the limits that apply to your key.HyperAPI runs on Hyperbots' finance-specific models, including HyperLM and in-house vision-language models.
Adapting LLMs and VLMs to invoice extraction without human-annotated training data, using synthetic data generators (InvoicePy and TemplatePy). Accepted at the 4th Deployable AI Workshop, AAAI 2026.
A synthetic invoice set with HTML, OCR-style text and gold labels in the Extract output format. Use it to benchmark HyperAPI or your own pipeline.
Posts on deterministic extraction from financial PDFs, schema-as-contract, and classifying more than 1,400 bank statement templates.