Open AI datasets, curated for real model work.
A searchable directory of public datasets for language models, vision systems, multimodal training, retrieval, evaluation, legal AI, healthcare AI, and public sector research. Filter by category and age, then jump straight to the original source.
Search, filter, and scan by use case.
Categories are practical rather than academic: the point is to help people find useful dataset starting points fast.
No datasets matched.
Try a broader search term or reset the filters.
How this list is curated
We prioritize reputable public sources, widely used benchmark collections, government portals, and well-known open data projects over random mirrors and SEO bait.
How age works
Age buckets are meant to help with discovery. Some older datasets are still foundational, so "evergreen" can matter more than "new."
Need a sovereign version?
If you need compliant ingestion, redaction, retrieval, evaluation, or hosting on Canadian infrastructure, talk to Anitarian.