As AI models become increasingly commoditized, startups are racing to build the software layer that sits on top of them. The place has an interesting interior. Osirisan open-source, Apple-only LLM server that lets users move between different native AI models locally or in the cloud, while keeping their files and tools on their own hardware.

Osiris evolved from an idea. Desktop AI companionDinoki, the co-founder of Osaurus Terence P Described as an “AI powered clip”. Danuki’s users have asked why they should buy the app if they still have to pay for tokens — the units AI companies charge for processing signals and generating responses.

This forced Pae to think more deeply about running AI natively.

“That’s how Osaurus started,” Pae, formerly a software engineer at Tesla and Netflix, told TechCrunch on a call. The idea, he explained, was to try to run an AI assistant natively. “You can do a lot of things natively on your Mac, like browsing your files, accessing your browser, accessing your system configurations. I thought this would be a great way to position Osaurus as a personal AI for individuals.”

Pae began making the tool publicly. An open source projectAdding features and fixing bugs along the way.

Today, Osaurus can connect flexibly with locally hosted AI models or with cloud providers like OpenAI and Anthropic. Users can freely choose which AI models they are using and keep other aspects of the AI ​​experience on their own hardware, such as the models’ own memory, or their files and tools.

Given that different AI models have different strengths, the advantage of this system is that users can switch to the AI ​​model that suits their needs.

Such a framework creates what Osorius calls a “harness” – a control layer that connects different AI models, tools and workflows through a single interface, similar to OpenClaw or Tools. Hermes. The difference, however, is that such tools are often aimed at developers who know their way around the Terminal. And sometimes, as in the case of OpenClaw, they can lead to security issues and problems.

Osaurus, meanwhile, offers an easy-to-use interface that users can use and mitigate security concerns by running things in a hardware-isolated, virtual sandbox. It limits the AI ​​to a specific scope, keeping your computer and data safe.

Of course, the practice of running AI models on your machine is still in its early days, as it is highly resource-intensive and hardware-dependent. To run native models, your system will need at least 64GB of RAM. To run larger models, such as DeepSeek v4, Pae recommends systems with around 128GB of RAM.

But Pae believes the need for native AI will diminish over time.

“I can see its potential, because intelligence per watt — which is like a metric for native AI — is growing significantly. It’s on its innovation curve. Last year, native AI could barely complete sentences, but today it can actually run tools, write code, access your browser, and order stuff from Amazon,” he said … and it’s getting better.

Osaurus today can run MiniMax M2.5, Gemma 4, Qwen3.6, GPT-OSS, Llama, DeepSeek V4, and other models. It also supports Apple’s on-device foundation models, Liquid AI’s LFM family of on-device models, and in the cloud, it can connect to OpenAI, Anthropic, Gemini, xAI/Grok, Venice AI, OpenRouter, Ollama, and LM Studio.

As a full MCP (Model Context Protocol) server, you can grant access to your tools to any MCP-compliant client. In addition, it ships with over 20 native plugins for Mail, Calendar, Vision, macOS Usage, XLSX, PPTX, Browser, Music, Git, File System, Search, Retrieve, and more.

More recently, Osorus was updated to include voice capabilities as well.

Since the project went live nearly a year ago, it has been downloaded north of 112,000 times, according to its website. The app competes with other tools that let you run models locally, e.g Allama, Mrs, LM Studioand others, but offers a distinctive feature set and presents itself as a more user-friendly option even for non-developers.

Currently, Osorus’ founders (including co-founder Sam Yu) are participating in the New York-based Startup Accelerator Alliance. They’re also thinking about next steps, which could see Osaurus being offered to businesses, such as in the legal space or healthcare, where running native LLMs could address privacy concerns.

As local AI models grow in power, the team believes this could reduce demand for AI data centers.

“We’re seeing this explosive growth in the AI ​​space where [cloud AI providers] There’s going to be scaling using data centers and infrastructure, but we feel like people haven’t seen the value of local AI yet,” Pae said. “Instead of relying on the cloud, they can actually deploy Mac Studio on-prem, and it should use a lot less power. You still have the capabilities of the cloud, but you won’t be dependent on the data center to run that AI,” he added.