PhotoPrism: AI-Powered Photo Library Manager
PhotoPrism organizes a photo collection and makes it searchable using machine learning that runs on your own hardware.
What it does well
Automatic classification tags photos by content, so searching for “beach” or “dog” works without you labelling anything. Face recognition groups people, places are mapped from EXIF coordinates, and RAW files are handled alongside JPEGs. Duplicate detection and quality filtering keep the library from filling with near-identical bursts.
It is notably stable and predictable compared with newer alternatives, and it runs acceptably on a NAS or a small server.
Versus Immich
This is the comparison people actually want. Immich has better mobile apps and far stronger automatic phone backup, which for many people is the deciding feature. PhotoPrism is more mature, changes less abruptly, and treats an existing organized photo folder as the source of truth rather than importing into its own structure.
If your photos already live in a directory tree you maintain, PhotoPrism fits that model better. If you want phone-to-server backup that behaves like Google Photos, Immich does that better today.
Storage model
PhotoPrism reads an originals directory and writes thumbnails and its index separately, leaving your files untouched and organized however you already had them. That is a meaningful difference: you can stop using PhotoPrism tomorrow and your library is exactly as it was.
Sponsorship and features
Some features are gated behind sponsorship of the project. The core library, search, and recognition are available in the free build.
License
PhotoPrism is released under the GNU Affero General Public License v3.0.