doc.cheap

Changelog: September 2026

Changes to the API and to this documentation published in September 2026, newest first.

2026-09-29 – 1.47.0, the history pages and a scan can be deleted

GET /v1/scans now answers a page at a time. limit sets the page size, from 1 to 100, and cursor continues from the next_cursor of the previous page, which is null on the last one. With neither, the first page is the same 100 rows the list returned before.

DELETE /v1/scans/{id} deletes one stored scan before its retention window ends: the result, its history row and its thumbnail. It cannot be undone, and the credit the scan drew is not returned.

The MCP server gains three tools on top of these: list_scans, get_scan and delete_scan.

2026-09-24 – 1.29.2, a damaged image is refused, not a failure

An image that starts like a JPEG or PNG and cannot be decoded, such as a file cut off after its header, is refused with 415 unsupported_media_type instead of 500 internal_error. Nothing is charged, and the sandbox key's free attempt is handed back.

GET /v1/scans now has its own reference page, like the other three endpoints, and is listed in llms.txt. The markdown copies of the pages and llms-full.txt carry their links as full addresses, so a link followed from one of them lands on its page. When the public sandbox key runs out of requests for the hour, the refusal now says so: Sandbox limit of 10 requests per hour per IP reached.

2026-09-24 – 1.28.0, an image that is not one is refused

POST /v1/scans reads JPEG and PNG. An image whose bytes are neither – a PDF, a HEIC photo, a text file, a file cut off after its first bytes – is now refused with 415 unsupported_media_type before recognition runs. It used to reach recognition and come back as 500 internal_error. Nothing is charged, and no free attempt of the sandbox key is spent.

2026-09-17 – 1.0.0, the first production release

The document-recognition API, at version 1.

Recognition. POST /v1/scans takes one image as base64 and returns the extracted data in the same response. The call is synchronous: no job to poll, no callback to register. Passports, identity cards, travel documents and driving licences are read off the printed page, the machine-readable zone and a barcode where one is printed.

One response shape. Eight groups – meta, document, holder, fields, mrz, images, quality and authenticity – described key by key on the response. document carries the document's kind, issuing state, number, series, date of issue and date of expiry, and whether it has expired. Every key is present; an absent value is null and an absent collection is empty. meta.schema_version is "1.0".

Fields in every script. Every field the engine reads is re-keyed to a stable vocabulary and published once per language, resolved from 418 assigned language identifiers. A document printed in two scripts carries both spellings, neither chosen for you.

The machine-readable zone, verbatim. mrz is a verdict with the zone beside it: the lines exactly as read, and one unbroken text a check-digit routine takes unchanged.

Result images. Seven crops, each capped by height – 250 px for the document crop, 100 px for the rest – returned by the call that produced them and never stored.

Billing. One credit is one US cent, charged only for a recognition that produced data. A credit is reserved before the engine runs and settled after. GET /v1/usage reports the balance and the period's counters.

Retention is the caller's choice. retain_hours decides how long a result stays readable through GET /v1/scans/{id}; 0 writes nothing down at all.

Safe retries. An Idempotency-Key on a scan request returns the first result instead of running and charging again.

Errors. One error shape everywhere, 21 stable codes, each with a page of its own at /errors/<code> that every error body links to.

Three ways in. The public sandbox key sk_sandbox_public recognizes real documents without an account; a registered account's own sandbox key answers from a fixed specimen; a live key bills.