The Labelgrade open dataset
Every grade we publish, free to download and reuse. 192 branded packaged foods with their independent 0–100 nutrition scores and letter grades, plus 74 common whole foods with USDA-referenced protein and macros. Released under CC BY 4.0 — use it anywhere, just credit Labelgrade with a link. Last updated 2026-06-04.
Download
| File | Format | Contents |
|---|---|---|
| Graded products | CSV | 192 branded products |
| Graded products | JSON | 192 branded products |
| Whole foods | CSV | 74 USDA reference foods |
| Whole foods | JSON | 74 USDA reference foods |
| Everything combined | JSON | products + foods + metadata |
What's inside
Graded products — one row per branded product, with identity (brand, name, UPC, USDA FDC id), per-serving nutrition, and the Labelgrade score:
slug, brand, product_name, category— identityupc, usda_fdc_id— external identifiers for joining to other datasetsserving_size, calories, protein_g, total_fat_g, saturated_fat_g, total_carbs_g, fiber_g, sugars_g, added_sugars_g, sodium_mg— nutrition per servingscore(0–100) andgrade(A+…F) — the overall Labelgradeprotein_density_score, ingredient_quality_score, sugar_load_score, sodium_load_score, fiber_score, saturated_fat_score— the six dimension sub-scores (0–100)last_verified, report_url— provenance + a link to the full report
Whole foods — one row per generic food, per 100 g, with a common portion and USDA source:
slug, food_name, common_portion, portion_gramsprotein_g_per_100g, calories_per_100g, total_fat_g_per_100g, saturated_fat_g_per_100g, total_carbs_g_per_100g, fiber_g_per_100g, sugars_g_per_100g, sodium_mg_per_100gusda_fdc_id, usda_data_type, last_verified, report_url
License & how to cite
This dataset is licensed Creative Commons Attribution 4.0 (CC BY 4.0). You're free to share and adapt it for any purpose, including commercially, as long as you give credit. Suggested attribution:
Nutrition grades from Labelgrade (labelgrade.com), CC BY 4.0.
Nutrition figures are sourced from USDA FoodData Central and manufacturer labels; scores are computed by the transparent Labelgrade methodology. Grades reflect our editorial assessment at the time of verification — always check the current product label. The dataset is regenerated as we add and re-verify entries.