Reef AI: ask for any data. Try it
Looking for the overview — what this API returns, what it costs, and a call you can run without a key? See the UPCitemdb API page →
docs / upcitemdb

UPCitemdb

Products by UPC, EAN or ISBN from UPCitemdb: title, brand, model, MPN, images, ASIN, book author and the retailer offers it recorded, with price and date.

base /upcitemdb/v15 endpoints
post/upcitemdb/v1/lookup1 credit

Barcode → the UPCitemdb product record: title, brand, model and MPN, category, description, images, colour/size, package weight as an explicit {weight, weight_unit} pair (with weight_grams, weight_source and a weight_confidence flag), ASIN, author for books, and the retailer offers UPCitemdb lists (merchant, price, updated date) with the lowest and highest price; plus check-digit validation, GTIN-13 normalisation and the GS1 issuing country. A barcode UPCitemdb does not hold answers NOT_FOUND with the decoded code in error.detail.

ParameterAllowed / rangeDescription
barcoderequired—The barcode to look up: UPC-A (12), EAN-13 (13), EAN-8 (8), GTIN-14, or ISBN-13 (978/979…). Spaces/dashes are stripped; check digit is validated (mod-10). Aliases: upc, ean, gtin, isbn, code.
deep_weight = falseoptional—Spend ONE extra UPCitemdb call per barcode to read the weight from UPCitemdb's JSON row when the product page printed none. Measured on 61 retailer-sourced barcodes: weight coverage 19.7% off (default, 1 call) → 29.5% on (2 calls). Read weight_confidence before trusting it.
Try in playground →
post/upcitemdb/v1/batch1 credit

Up to 20 barcodes in one call → one UPCitemdb record per barcode, in input order, each with found/checked flags; a barcode with no record carries its decoded identifier instead. Billed per barcode found.

ParameterAllowed / rangeDescription
barcodesrequired—Up to 20 barcodes — a JSON array or a comma-separated string (competitor caps batch at 10; we allow 20). Each is normalized + validated; results preserve input order with per-item source.
deep_weight = falseoptional—Spend ONE extra UPCitemdb call per barcode to read the weight from UPCitemdb's JSON row when the product page printed none. Measured on 61 retailer-sourced barcodes: weight coverage 19.7% off (default, 1 call) → 29.5% on (2 calls). Read weight_confidence before trusting it.
Try in playground →
post/upcitemdb/v1/by_mpn1 credit

Manufacturer part number or model → the UPCitemdb products carrying it, exact model/MPN matches first (each row says matched_mpn and exact_match).

ParameterAllowed / rangeDescription
mpnrequired—Manufacturer Part Number / model number → matching products. Aliases: model, model_number, part_number. Results whose model/mpn field equals the query are surfaced first (exact_match=true).
page = 1optional1–10001-based page number (clamped 1-1000). Page forward with meta.next_page. Note: Open Food Facts throttles deep pages, so a deep page may take an extra moment while the engine retries.
page_size = 20optional1–50Results per page (1-50, default 20); out-of-range values are clamped.
Try in playground →
post/upcitemdb/v1/validate1 credit

Validate and decode a barcode with no network call: mod-10 check digit, GTIN type, GTIN-13 form, GS1 issuing country, ISBN detection.

ParameterAllowed / rangeDescription
barcoderequired—The barcode to look up: UPC-A (12), EAN-13 (13), EAN-8 (8), GTIN-14, or ISBN-13 (978/979…). Spaces/dashes are stripped; check digit is validated (mod-10). Aliases: upc, ean, gtin, isbn, code.
Try in playground →
Built for volume
5M+ requests a day

Measured at 60 requests a second across the fleet, with no central bottleneck. Volume pricing is on request, and per-key limits are raised for high-volume accounts.

Missing a source?
We build it

Tell us a site we do not cover yet and it becomes an engine. A customer asked for bestprice.gr on a Sunday and it was in the catalog the next day.

Support
2 minute median reply

Median time from a question in the live chat to the first answer, measured across every answered conversation. Setup help included, no support tier to buy.

One key, one balance
Every API included

No per-site plans and no separate subscriptions. One key and one credit pool across the whole catalog, so adding a source costs nothing up front.

Planning something large? Tell us the volume and the sources and we will come back with what it costs and what we would have to build.