HN: AI File Sorter 1.3 – Add your own Local LLM for offline file organization (github.com)

AI File Sorter 1.3.0 has been released for Windows, macOS, and Linux (available at filesorter.app and SourceForge). The update adds the ability to plug in your own local LLMs so file categorization can run entirely offline, along with richer categorization options, support for multiple categorization and interface languages, and optional categorization whitelists. These changes make the app more customizable—users can define categories, restrict processing to specified folders or file types, and run models locally for privacy and latency benefits. The full changelog is available from the project page. For the AI/ML community this is significant because it lowers the barrier to on-device, model-driven file management without sending data to the cloud. Allowing users to attach their own local models means practitioners can test different models and prompts, benchmark performance and accuracy in real-world file-organizing tasks, and maintain data governance. Technical implications include trade-offs between model size/performance and local resource use, easier multilingual workflows via localized interfaces and categorization languages, and improved control through whitelists—useful for automated workflows, data labeling, and secure enterprise deployments.