Molten: Local AI Chat

关于这款应用

Molnet enables users to run large language models entirely on their devices, ensuring that personal data never leaves the hardware. The application supports connections to local servers like Ollama and Swama (optimized for Apple Silicon), as well as built-in on-device models. This architecture eliminates cloud dependencies, telemetry, and tracking, providing a fully private experience.

The app features a native SwiftUI interface available on both iOS and macOS, offering dark mode, markdown rendering, syntax highlighting, and keyboard shortcuts. On macOS, a floating panel allows for quick access to chats. Users can switch between self-hosted servers and model selections seamlessly. Real-time streaming responses are accompanied by performance analytics, displaying metrics such as tokens per second and evaluation rates. The software is open-source on GitHub, allowing for full code audits. Recent updates have improved accessibility with VoiceOver labels, fixed connection stability issues, and enhanced privacy settings by masking API keys.