Osaurus brings local and cloud AI management to Mac

Osaurus, an open-source, Apple-only Large Language Model (LLM) server, has launched to help users manage both local and cloud-based artificial intelligence models. The tool functions as a “harness”—a control layer that connects different AI models, tools, and workflows through a single interface. Since the project went live nearly a year ago, it has been downloaded north of 112,000 times.

The software evolved from a predecessor desktop AI companion called Dinoki. Co-founders Terence Pae and Sam Yoo, who are currently participating in the New York-based startup accelerator Alliance, designed Osaurus to allow users to switch between local and cloud AI models while keeping their files and tools on their own hardware.

Running local AI models is highly resource-intensive and hardware-dependent. To run local models on Osaurus, a user’s system requires at least 64GB of RAM, while larger models like DeepSeek require about 128GB of RAM. Despite these heavy hardware demands, Osaurus aims to provide an easy-to-use interface, contrasting with developer-focused tools like OpenClaw, Hermes, Ollama, Msty, and LM Studio. Additionally, while tools like OpenClaw may pose security issues and holes to worry about, Osaurus addresses security concerns by running operations inside a hardware-isolated, virtual sandbox to protect user data and keep the system secure.

The platform also functions as a Model Context Protocol (MCP) server—a protocol allowing clients to access tools—and ships with over 20 native plug-ins for applications like Mail, Calendar, and the filesystem.

While local AI currently demands significant hardware, Pae believes local AI’s needs will come down in time. “I can see the potential of it, because the intelligence per wattage — which is like the metric for local AI — has been going up significantly. It’s on its own curve of innovation. Last year, local AI could barely finish sentences, but today it can actually run tools, write code, access your browser, and order stuff from Amazon … It’s just getting better and better,” Pae said.

The founders are also exploring next steps, including offering Osaurus to businesses in privacy-sensitive sectors like legal and healthcare. By deploying a Mac Studio on-prem—meaning on-premises hardware located at the user’s site—businesses could run capable AI systems with substantially less power, reducing their dependence on massive cloud data centers.

Why it matters

Osaurus provides a control layer for users to manage both local and cloud AI models on Apple hardware, aiming to address privacy concerns and potentially reduce the need for large-scale AI data centers. By shifting processing to local hardware, the startup highlights a path toward energy-efficient, private AI deployment that bypasses traditional cloud infrastructure.