SweepDisk

SweepDisk requires an Apple Silicon Mac (M1/M2/M3/M4) running macOS 14 Sonoma or later. The local AI model leverages Apple Silicon's Neural Engine for acceleration.

On first launch, approximately 2GB of model files are downloaded. They're stored in the app support directory and don't need to be downloaded again. Runtime memory usage is about 1-2GB.

No. All operations require your confirmation before executing. Classification and rename suggestions are previewed first — you can confirm individually or in batch. File moves never delete originals, and undo is supported.

Apart from the initial AI model download, optional model upgrades, and checking for app updates, SweepDisk runs completely offline. All file analysis happens locally — your data never leaves your computer.

SweepDisk is an indie project. We believe great tools should be accessible to everyone. We may introduce premium features in the future, but core organizing functionality will always be free.

macOS Clean Up only arranges icons by type or date with fixed rules. SweepDisk understands file content, classifies semantically (distinguishes invoices from papers), renames intelligently, and proactively suggests cleaning unnecessary files.