Enterprise-grade training data for food-aware AI. Clean labels, global coverage, continuous refresh — the substrate your models are missing.
12M+ restaurants with identity, location, cuisine, hours, price band and menu pointers across 180+ countries.
Structured menus with sections, modifiers and descriptions; canonical dish labels with multilingual aliases for training NLU and retrieval.
Dish-to-ingredient mappings with confidence, allergen propagation, dietary labels and substitution graphs for reasoning models.
Dish-level calories and macros traced to ingredients, with confidence bands — built for health-aware assistants and dietary apps.
Historical price trajectories per dish and location — train forecasting, market intelligence and dynamic-pricing models.
Precomputed embeddings of restaurants, dishes and ingredients in the knowledge graph — ready to drop into your vector store.
Every dataset ships with clear, model-training-friendly licensing and provenance. No scraped-dark-corner surprises.
Snapshots publish monthly; streaming deltas daily. Your training corpus never goes stale.
Datasets deliver to your cloud (S3, GCS, Azure) in your VPC. Nothing is shared with other customers.
Tell us what you're training. We'll send sample slices and a licensing quote within two business days.
