Clean Downloads
Sort installers, receipts, exports, PDFs, photos, screenshots, and temporary files into folders you can scan later.
Feature overview
AI File Sorter features
A fuller look at the controls behind the download page: review-first cleanup, local model options, safer recursive scans, undo, and privacy-focused workflows.
Tired of digging through cluttered folders with meaningless file names? Let AI File Sorter help you clean Downloads, rename photos and scanned documents, organize screenshots, and review old archive folders before anything changes.
Choose a focused cleanup flow, inspect the proposed categories and filenames, then apply only the rows you trust. Dry run, file previews, undo, and project-folder protection keep everyday cleanup reversible and easier to verify.
Use it with local folders, synced cloud folders, external drives, and mounted NAS shares. Local processing can stay fully offline, or you can bring your own OpenAI, Gemini, or compatible endpoint when you want cloud model power.
Common cleanup jobs
Rename photos and scanned documents
Use image and document content to suggest clearer names for camera exports, scans, PDFs, Office files, and text documents.
Organize screenshots
Group screenshots and UI captures by visible content, project, or reference type.
Sort NAS archives
Point AI File Sorter at mounted network shares or external backups and review the plan before moving files.
What the app helps you control
-
Guided cleanup flows for common file piles
Use focused starting points for jobs like Clean Downloads, Rename photos and scans, Organize screenshots, or Sort archives. The app shows a review plan first, so you can approve only the changes you trust.
-
local visual image analysis for picture files, with Gemma 3 4B IT as the default backend
The app can use local visual models to understand images, handle screenshots and UI captures more intelligently, and suggest useful categories or filenames without uploading pictures.
-
Smarter category whitelists with category-specific subcategories
Create focused category lists where each main folder can have its own allowed subfolders, so a Documents cleanup and an Images cleanup do not have to share the same subcategory rules.
Show more
-
File previews in the review window
Preview files while checking suggested categories and names, making it easier to approve the right changes without jumping back and forth between windows.
-
Project-folder protection for safer recursive scans
When scanning subfolders, AI File Sorter avoids reorganizing recognized Unity, Unreal, Godot, Blender, Git, and common source-code project folders where moving files could break a project.
-
Document content analysis for PDFs, Office files, and common text formats
Instead of relying only on filenames, AI File Sorter can read supported documents to make better sorting and naming suggestions.
-
Local learning from approved reviews plus cache and reset tools
Approved category decisions can be reused as local hints in future runs, and Settings includes tools to clear caches or reset learned behavior when you want a clean slate.
-
Rename-only flows with suggested filenames and clear Renamed / Renamed & Moved status labels
Use rename-only mode when you want clearer filenames without moving files, with review labels that show exactly what changed.
-
Optional metadata-based filename suggestions for supported audio and video files
For media files, available metadata can help create more meaningful filenames before you approve the changes.
-
Picture-only or document-only processing toggles to focus runs
Limit a run to pictures or documents when you want faster, more targeted cleanup for a specific file type.
-
Optional image creation-date suffixes for time-aware categories
When image dates are available, categories can include date context so photo folders are easier to scan later.
-
System compatibility check that helps pick the best LLM for your hardware
The app checks your machine and suggests model choices that fit your available hardware instead of leaving you to guess.
-
Custom API endpoints in the LLM selector for local servers or self-hosted gateways
Point the app at your own compatible endpoint when you run models locally or through a private gateway.
-
Progress is saved as you go so long runs are resilient to interruptions
Long sorting jobs save progress during processing, reducing the risk of losing work if something interrupts the run.
-
Preview-only dry run with a From/To table so nothing moves until you say so
Review the proposed source and destination paths first, then decide whether to apply the moves.
-
Persistent Undo: revert the latest sort even after closing, thanks to saved plan files
Undo information is saved to disk, so you can revert the last applied sort even after closing the app.
-
Bring your own OpenAI or Gemini API key for remote LLMs; pick any ChatGPT or Gemini model
Use cloud models only with your own keys, giving you control over provider, model choice, and account usage.
-
Folder integrations for cloud sync, external drives, and NAS shares
Use folders already available to your operating system, including Dropbox, Google Drive, OneDrive, iCloud Drive, USB drives, SMB shares, and other mounted NAS locations.
-
Add custom local text and visual models, and choose where model files are stored
Download supported local models quickly, reuse Gemma 3 4B IT for both text and visual workflows when applicable, add your own local model files, and keep model downloads in the storage location you prefer.
-
Switch between More Refined and More Consistent categorization modes
Choose More Refined when you want more specific top-level folders, or More Consistent when you want similar files to stay under steadier broad categories.
-
Better accessibility and clearer progress updates
Improved labels and progress announcements make the app easier to follow with screen readers and during longer analysis runs.
-
More interface languages plus localized Quick Start help and an in-app FAQ link
Use localized interface text and get guided help faster with built-in Quick Start and FAQ entry points across the expanded language set.
-
GPU acceleration via CUDA, Vulkan, and Metal with smarter backend selection
Compatible NVIDIA, AMD, Intel, and Apple hardware can speed up local model processing, and recent releases improve backend choice and startup reliability.
-
Secure and private: your data stays on your machine
Local processing keeps files and model analysis on your computer unless you explicitly choose a remote provider.
-
Cross-platform and open source: available for Windows, macOS, and Linux (see GitHub)
The app is available across major desktop platforms, and the source code can be inspected or contributed to on GitHub.