Capability · OCR

Turn documents into structured markdown.

Extract text and layout from scanned forms, faxes and PDFs — page-billed, PHI-eligible via Azure, and resident in Switzerland, the EU or the US.

POST /v1/ocr

What you get

Built for regulated workloads

OCR is the one endpoint that is not OpenAI-shaped — OpenAI has no OCR equivalent — so /v1/ocr uses the Mistral-native request (a document plus optional page range and table format) and returns per-page markdown. It is available in Switzerland, the EU and the US; elsewhere it returns no_route.

Output

Per-page markdown

Each page comes back as clean markdown with structure preserved, plus a page count and the raw provider payload — ready to chunk, index or feed into a chat call.
PHI

PHI on Azure

PHI OCR runs on Azure Document Intelligence (CH / EU / US) under the verified HIPAA BAA. EU general OCR can also use Mistral OCR.
Layout

Tables & structure

Request a table format to keep tabular data intact, or a page range to OCR only part of a large document.
Residency

Resident extraction

Documents are processed in the region you pin and never persisted by the proxy. CH defaults to Azure; EU defaults to Mistral OCR with Azure for PHI.

Availability & pricing

Where it runs, what it costs

Every route is region-resident and the PHI gate is enforced per call. Prices include the flat +10% gateway margin and mirror the live /v1/pricing rate card.

RegionProviderModelTierPrice
CHAzure Doc IntelligenceOCRPHI$0.00165 / page
EUMistral OCROCRGeneral$0.0022 / page
EUAzure Doc IntelligenceOCRPHI$0.00165 / page
USAzure Doc IntelligenceOCRPHI$0.00165 / page
OtherOCRNo routeCH/EU/US only

Billed per page (Mistral OCR with annotations is priced higher). Gateway margin included. Mistral OCR carries general traffic only; PHI OCR is Azure.

Try it

A real call, end to end

OCR a scanned referral in Switzerland; get markdown back, page by page.

curl
curl https://phi-cloud.com/api/v1/ocr \
  -H "Authorization: Bearer $PHI_API_KEY" \
  -H "X-Region: CH" \
  -H "X-PHI: true" \
  -H "Content-Type: application/json" \
  -d '{ "document": "data:application/pdf;base64,JVBER..." }'
response
x-phi-routed: azure-docintel-ocr/CH/phi
x-phi-tier: phi
{ "pages": [{ "index": 0, "markdown": "# Referral..." }],
  "pageCount": 3 }

Residency & coverage

  • CH OCR → Azure Document Intelligence (Switzerland), PHI-eligible.
  • EU OCR → Mistral OCR (general) or Azure (PHI).
  • US OCR → Azure, PHI-eligible.
  • Every other region → no_route. OCR is CH/EU/US only.

FAQ

Common questions

OpenAI has no OCR endpoint, so there is no shape to match. /v1/ocr uses the Mistral-native request and returns per-page markdown — the closest thing to a standard in the OCR space.
Azure Document Intelligence, under the verified HIPAA BAA, in CH, EU and US. Mistral OCR is general-traffic only (its DPA does not cover special-category health data).

Ready when you are

Put document ocr in production — without giving up your data.

Spin up a key in minutes. The residency and PHI posture above arrives unchanged.

Free to test · Prepaid credits, no subscription · No data retained