Tabscanner/API v2
Dashboard

Changelog

Changes to API v2 and to this documentation, newest first. Within v2 all changes are additive; see Versioning.

harajuku for every account

  • model: "harajuku" is now available on every plan, including prepaid credit. features.models on GET /v2/account is always true.
  • The -pro models still need the Pro package; MODEL_NOT_PERMITTED now only applies to them.

Pro models

  • New model value harajuku-pro, a more thorough version of harajuku, available with the Pro package. See Models.
  • features.proModels on GET /v2/account.

API v2 launched

A new REST interface to the same receipt processing, at https://api3.tabscanner.com/v2. The original API keeps working unchanged and is documented at docs.tabscanner.com.

  • Resource-style endpoints: POST /v2/receipts, GET /v2/receipts/{token}, GET /v2/account, GET /v2/regions, GET /v2/custom-fields.
  • Standard HTTP status codes, including 202 Accepted with Location and Retry-After for uploads.
  • Errors as RFC 9457 problem details with stable, documented code values and a requestId.
  • API keys sent as Authorization: Bearer (or the apikey header), never in URLs.
  • Idempotency-Key for safe upload retries that are never charged twice.
  • Retry-After on pending results to pace polling.
  • Tabscanner model processing with model=harajuku.
  • Up to three images per upload for accounts with multi-image uploads enabled.
  • X-Request-Id on every response.
  • Documentation: OpenAPI 3.1 generated from the API itself, JSON Schema for the result object, llms.txt, llms-full.txt, a page for AI agents, and code samples in eight languages, each run against the API by an automated test.