The data layer forhow India eats.
Nutrition, recipes, packaged products and regional diets, measured in Indian kitchens and labs. Clean, versioned and ready for your app, your research or your model.
No spam. Unsubscribe anytime. Cancel anytime. Open data for public health; paid for products.
Built on NIN IFCT 201727,904 Indian labels84,312 recipes in 14 languages36 states & UTs
Kanda poha · Maharashtrav2026.10
254 kcal
per plate · 180 g median
Glycaemic load, measuredCGM
| Rice, dal, curd | GL 21 |
| Jowar bhakri, pitla | GL 14 |
| Idli, sambar | GL 22 |
Masala oats, 38 glabel v3
1,180 mg sodium
↓ 12% since 2025 reformulation
Tamil Nadu · typical dinner
21:40 · rice, rasam, poriyal
Six datasets, one schema
Every record joinsevery other.
Shared IDs, units and provenance fields, so a recipe links to its ingredients, its packaged equivalents and the regions that cook it.
Indian household units
Nobody in India eats "100 g".
People eat a katori of dal, two phulkas and a ladle of sabzi. Kosha weighs real household portions across regions, so your app can ask the way people answer.
Dish
Portion
150g
Median of 1,920 weighed portions across 11 states · Kosha field survey 2025–26
Explore a sample
3,200 records,sorted four ways.
Each square is one record. Re-sort them by region, by glycaemic load or by protein, and hover any square to read it.
API
Speaks pohe,aval and अवलक्की.
Search that understands names
pohe, aval and अवलक्की resolve to the same flattened rice, ranked by your user's state.
Provenance on every number
Lab-measured, recipe-calculated or label-read, with sample size and error. You decide what counts as good enough.
Versioned like code
Monthly releases with changelogs. Pin a version for a study, or follow the latest for your app.
Pricing
Open for public health.Paid for products.
Core
Free · CC BY 4.0
- 1,000 common foods, 38 nutrients
- Household portions, 11 states
- CSV and Parquet
Builder
₹40,000 / month
- All six datasets via API
- 1 million calls a month
- Barcode and name search
- Monthly versioned releases
Enterprise
Custom
- Bulk exports and on-prem
- Lab studies on your products
- Reformulation benchmarks
- A dedicated data scientist
What are youbuilding?
Tell us in a sentence. We review every request within two working days and point you to the right dataset.