Frequently Asked Questions

A normalized dataset and, later, a REST API of packaged food and beverage products, built for developers. Every record is keyed by a validated GTIN-14 barcode and carries the facts printed on the label in structured form: the full ingredient statement and its parsed tree in exact label order, nutrition facts normalized to name, amount, unit and % daily value against the stated serving size, declared and precautionary allergens as normalized identities, dietary claims, sweeteners, caffeine and alcohol flags, package warnings, categories, and product attributes like storage condition and net content. It is built for product teams who would otherwise spend months collecting and cleaning this data themselves: app builders, e-commerce platforms, and research teams.

The homepage shows a complete, real record from the sample. In short: identity (id, GTIN-14, name, brand, product type), classification (category path, storage condition, physical form, flavor, net content, pack count), the ingredient statement verbatim plus the parsed tree with per-ingredient thresholds where the label states them (for example "contains 2% or less of"), declared and may-contain allergens with the verbatim statement, dietary claims and an organic flag, the nutrition panel with serving size, basis and preparation, sweeteners, caffeine and alcohol fields, package warnings, directions, countries of sale, and the file names of the main product photo and any label photos.

500,000+ products at launch. The free sample holds 1,000 of them across 520 brands, and every record in the full catalog uses the same schema and file formats as the sample, so what you build against the sample carries over unchanged.

The catalog is built from publicly available product and label information, normalized by our own pipeline: barcode validation, ingredient parsing, nutrition normalization, allergen tagging and category cleanup. Label photos are read to fill gaps and to check extracted values against the package. The normalization layer is what makes the data usable in an application: we preserve what the label states and never estimate or invent values.

1,000 products across 520 brands, delivered as products.jsonl, products.csv and products.sqlite (identical records, pick your format), a SCHEMA.md documenting every field, and an images folder with 1,286 photos: main product images plus label photos (ingredient list, nutrition panel, back of pack).

Enter your email on the homepage. We send a personal download link right away; the page it opens hands you the zip directly. The link is signed to your address, so please don't forward it: your colleagues can request their own in ten seconds.

Evaluate it freely: load it, query it, prototype against it, build internal demos, benchmark it against other sources. What you may not do is redistribute the files or their contents, publish them as a dataset, or use them in a shipping product or a public model without a license. Refer to the Terms of Service for full details.

Yes. The sample is rebuilt as the catalog grows, so the link you receive always points at the latest sample. Ids are permanent, so records can be diffed by id between versions.

By end 2026. Everyone who has requested the sample hears first; we send exactly one email when the full dataset is available.

Pricing is announced at launch. The plan is the same shape as our sister product, The Beauty API: one-time dataset licenses first for teams that want the raw files, with a hosted API on monthly plans to follow. Requesting the sample puts you on the launch list at no cost and with no obligation.

Two ways, in this order: downloadable dataset snapshots first, exactly like the sample (JSONL, CSV, SQLite, images, schema documentation), then a hosted REST API (JSON over HTTPS, API-key auth, barcode lookup as a first-class endpoint).

Typical uses: nutrition and diet-tracking apps that resolve a scanned barcode to full label facts; grocery e-commerce catalogs enriched with structured ingredients and allergen badges; allergy and dietary-preference filtering; health and fitness platforms logging real products; and research or ML teams who need clean, structured label data at scale.

We preserve what the label states and never estimate or invent values: if a label does not state a serving size or a nutrient, the field is null rather than guessed. Manufacturers reformulate products, so any label dataset can lag the shelf. Always verify against the physical package before relying on it for health-critical decisions. See the disclaimer.

The Food API is a product of The Direct API Pte. Ltd., a company incorporated in Singapore. It is the sister product of The Beauty API, built on the same pipeline architecture and the same one-time-license-then-API model.

Email support@thefoodapi.io. We are especially interested in what you are building, what you make of the sample, and which coverage you need; it shapes the roadmap.