Linger 0.5.1 · Chrome extension 0.5.3
A clearer way back to your images.
One search across quick find and Recall, with clearer settings.
Describe what you remember
Combine a scene, a color and a remembered source in one query, such as “a red dragon poster, maybe from Twitter.” With the optional local recall runtime configured and enabled, press Enter to search indexed images without tagging them first.
A remembered source is a preference. An explicit restriction such as “only from Twitter” limits results to matching saved source records. Missing source records are not guessed from the image.
One search, two places to work
Start with Command-F in the library, then continue in the Recall workspace with the same query, scope, results and page. Escape returns to browsing while keeping the search session. Add a memory, remove a clue or undo an edit as you recognize more of the image. A new search clears the session.
The similar-image buttons have been removed. Recall now centers on describing the image you want to find.
Optional candidate review
When you connect and enable your own model, Linger can send the search description, up to 24 local candidate previews and their saved records for review. Local candidates remain available if review fails or stops. The model is off by default; with local recall disabled, semantic search uses saved text.
Settings organized by task
Settings now has four sections: General, Appearance, Recall & AI, and Account & shortcuts. Preview the existing themes, check local indexing progress and open connection diagnostics when needed. Switching sections keeps unsaved model settings in place.
Availability
Linger is available for macOS with a 14-day full trial. Each account receives one trial starting with its first successful sign-in. No card is required, and there is no automatic charge.
Local recall remains experimental and off by default, with a separately configured Apple Silicon runtime. Model review checks a limited candidate set and can miss a target; retrieval quality across other libraries and large-library performance remain under evaluation. See current trial and license terms.