20,984 records · 1940s–2010s

Historical Ground Truth for the AI Era

Curated, copyright-cleared multimodal datasets built from historical archives. Scanned originals paired with structured extraction — ready for Vision-Language Model training.

Explore Datasets View Pricing

Data Rooms

Each room is a curated, copyright-cleared multimodal dataset ready for AI training and research.

Historical Recipes

20,000+ multimodal recipes from community cookbooks (1940s–2010s)

20,984Multimodal Recipes

Data Ingestion & Refinement

multimodalcopyright-cleared1940s–2010sstructured JSON

Coming Soon

Vinyl Record Metadata

10,000+ vinyl records with sleeve scans and structured metadata

10,000Vinyl Records

multimodalmusic1950s–1990s

Coming Soon

Technical Blueprints

1,000+ woodworking blueprints with dimensional extraction

1,000Technical Blueprints

multimodaltechnicaldimensional data

Coming Soon

Magazine Data Refineries

1914–1925 magazine pages with full-text & ad extraction

5,000Magazine Pages

multimodalhistorical1914–1925full-text