Osaurus: open-source LLM server for Mac bridges local and cloud AI models

Osaurus, an open-source LLM server for Apple devices, lets users switch between local and cloud AI models while keeping files and tools on their own hardware. Co-founder Terence Pae, a former software engineer at Tesla and Netflix, started the project after users of his earlier desktop AI companion Dinoki questioned paying for tokens. Osaurus evolved into a 'harness' that connects models like OpenAI, Anthropic, Gemini, and local options such as Llama and DeepSeek V4. It runs in a hardware-isolated sandbox for security and requires at least 64 GB of RAM for local models. The app has been downloaded over 112,000 times since launch nearly a year ago. Osaurus competes with Ollama, Msty, and LM Studio but targets non-developers with an easy-to-use interface. It supports over 20 native plugins for Mail, Calendar, Vision, and more, and recently added voice capabilities. Founders Pae and Sam Yoo are participating in the New York-based startup accelerator Alliance and exploring business applications in legal and healthcare sectors where privacy is critical. Pae believes local AI's efficiency per watt is improving rapidly and could reduce reliance on data centers.