Top Open Datasets
AI dataset and evaluation records ranked by usage, license clarity, provenance, trust signals, and downstream utility.
Focused lists make the catalog useful immediately while mirror hosting remains selective and reviewed.
- Multi-Source AI Popular non-Hugging Face records from ModelScope, GitHub, arXiv, OpenML, Zenodo, Dataverse, DataCite, and other indexed open-source AI sources.
- New Open Source Recently updated public open AI artifacts across every indexed source.
How this ranks. The order uses the recorded signals described above. Runtime and verification details live on each model’s record. Self-reported evidence has no trust or ranking impact.
Mirror candidate means the files still live upstream, not on our servers yet. Hosted files do not establish a complete runnable model. Download counts, where recorded, are downloads on Hugging Face.
Showing stale catalog ·
| Type | Model | Recorded size | Downloads | Updated |
|---|---|---|---|---|
opus ubuntuHelsinki-NLP/opus_ubuntuContent controls: not assessed.DetailsWhat it does: a dataset for translating text between languages.Source: Helsinki-NLP on Hugging Face.
| 1.62 MB | 396 downloads on Hugging Face | 2026-07-11 |