Cosmos
AI-powered local media search that finds your files by their content
Visit Product
172 upvotes
1,330 views
About
Cosmos is a local AI application that enables semantic search across your media files — photos, videos, and documents — by their visual or textual content rather than just filenames and metadata. Powered by CLIP and other AI vision models running locally, Cosmos understands what is in your photos and videos, making it possible to find images of 'sunset at the beach' or 'team meeting in a conference room' without any manual tagging.
All processing in Cosmos happens entirely on-device — no images or videos are uploaded to external servers. The AI indexes your local media library once and maintains a searchable semantic index that can query across thousands of files in milliseconds. This combination of privacy and capability addresses a genuine gap: cloud photo services offer AI search but require uploading your photos to their servers.
Cosmos is particularly valuable for photographers, videographers, and professionals with large local media archives who need to find specific content efficiently. Its local-first approach also makes it suitable for sensitive media that cannot be stored in cloud services.
All processing in Cosmos happens entirely on-device — no images or videos are uploaded to external servers. The AI indexes your local media library once and maintains a searchable semantic index that can query across thousands of files in milliseconds. This combination of privacy and capability addresses a genuine gap: cloud photo services offer AI search but require uploading your photos to their servers.
Cosmos is particularly valuable for photographers, videographers, and professionals with large local media archives who need to find specific content efficiently. Its local-first approach also makes it suitable for sensitive media that cannot be stored in cloud services.
Product Features
- Semantic search across photos and videos by content
- Completely local: no cloud upload required
- CLIP-powered visual understanding
- Natural language queries for media search
- Automatic library indexing and updating
- Support for major image and video formats
- Face grouping and recognition
- Location-based search from photo metadata
- Fast search across libraries of 100,000+ files
- macOS and Windows native application
- Completely local: no cloud upload required
- CLIP-powered visual understanding
- Natural language queries for media search
- Automatic library indexing and updating
- Support for major image and video formats
- Face grouping and recognition
- Location-based search from photo metadata
- Fast search across libraries of 100,000+ files
- macOS and Windows native application
About the Publisher
Cosmos was created to solve the fundamental problem of finding specific images and videos in large local libraries without cloud services. The project emerged from the availability of efficient open-source vision models that could run locally on consumer hardware, making it finally possible to offer the kind of AI-powered media search that cloud services provide — but without the privacy tradeoff. Cosmos serves photographers, creative professionals, and privacy-conscious users who maintain local media libraries.