OGAM is a mobile and desktop application that brings offline AI capabilities including chat, vision, speech-to-text, and image generation to phones and Macs without requiring cloud services or API keys.
The tool addresses the need to run AI models locally while maintaining complete privacy and avoiding dependency on external services. It accomplishes this by supporting GGUF-format language models, vision capabilities, Whisper speech-to-text, Stable Diffusion image generation, and tool calling—all executable on device hardware including CPUs, GPUs, and NPUs. The application also provides local-network server functionality, allowing models to be shared across devices on the same network without data leaving the user's infrastructure.
OGAM suits developers and users prioritizing privacy who want to deploy AI features on mobile devices or Macs without cloud connectivity or account management overhead. It works well for projects requiring offline-first operation, whether for privacy-sensitive applications, environments with limited internet access, or scenarios where latency and data residency matter. The tool is built with React Native, making it cross-platform across iOS, Android, and macOS from a shared codebase.
The project shows consistent development activity with regular commits addressing bug fixes, feature additions, and dependency updates. Work spans multiple areas including core model inference, platform-specific integrations for iOS and Android, UI improvements, and expansion of supported model types and capabilities. The maintainers actively respond to issues and incorporate user feedback into the roadmap. Development demonstrates attention to both breadth—supporting diverse AI modalities and hardware targets—and depth, with ongoing refinement of performance and user experience across platforms.
