Osaurus launches a Mac app where local and cloud AI models work together

Osaurus has released a macOS app that combines local and cloud AI models in a single interface. The core idea is that the agent's memory, files, keys and tools stay on the Mac itself, and the user connects to external models only when it is truly necessary.

How Osaurus works

Osaurus positions itself not as yet another chat with a model, but as a "harness": a control layer on top of different AI models. Users can run local models directly on Apple Silicon, and if they need more powerful generation or a specific provider, they can switch to cloud services such as OpenAI, Anthropic, Gemini or Grok. Much of the context never leaves for an external server: history, memory, plugins and working tools are stored on the device.

The project grew out of an earlier idea, the Dinoki desktop AI companion. According to co-founder Terence Pae, users asked why they should buy an app if they would still have to pay separately for model tokens. That pushed the team toward a local scenario: if the AI lives on your Mac, it can work with files, system settings and the browser like a personal assistant, rather than just another chat window.

What the app can do

Osaurus is betting on a hybrid approach: local models for privacy and control, cloud models for tasks that need maximum quality or speed. What matters here is not only the list of models, but that the whole workflow lives in one native macOS app: no terminal, no manual configuration and no scattered services. The product is trying to carve out a niche between developer-first tools like Ollama and more mainstream AI products that focus on a friendly interface.

• Switching between local and cloud models without changing your work environment
• Memory, chat history and keys stored on the Mac itself
• Support for MCP clients and access to tools through a single server
• More than 20 built-in plugins for mail, calendar, browser, Git, files and office formats
• Voice mode and an isolated sandbox environment for running code

According to the company, Osaurus already supports MiniMax, Gemma 4, Qwen 3.6, Llama, DeepSeek V4, Apple Foundation Models and the Liquid AI LFM family, and in the cloud it can connect to OpenAI, Anthropic, Gemini, xAI, OpenRouter, LM Studio and other providers. The project, launched about a year ago, has passed 112 thousand downloads.

Another important detail: the app works as a full MCP server, so it can be built into a broader stack of AI tools rather than used only as a standalone desktop client.

Where the weak spots are

The main limitation is plainly hardware. For running models locally, Osaurus recommends a system with at least 64 GB of RAM, and about 128 GB for large models on the level of DeepSeek V4. In other words, this is not yet about a base MacBook Air for the mass user, but about the Mac Studio, well-equipped MacBook Pro models and other machines with memory to spare. The platform is also aimed only at the Apple ecosystem.

The second problem is that local AI is still at an early stage. But this is where the founder sees fast progress: a year ago local models struggled to finish sentences, and now they can call tools, write code and work with a browser. Osaurus is trying to package that leap into a product for ordinary users, adding a virtual sandbox for safety and an interface that needs no terminal. Looking ahead, the team is also eyeing B2B scenarios, such as law firms and healthcare, where running models locally may reduce risks for private data.

What this means

The AI market is gradually shifting from a race between models to a fight for the control layer on top of them. Osaurus is betting that users need not just access to Claude or GPT, but a controlled environment where they can combine local and cloud models without losing memory, tools and privacy. If this approach catches on, the Mac could become not just a client for AI, but the main place where a personal agent actually lives.