Osaurus is reshaping how Mac users think about AI by letting them run powerful language models locally while maintaining the flexibility to tap into cloud services when needed. The open-source platform, which has been downloaded over 112,000 times since launching nearly a year ago, represents a growing shift away from the assumption that all AI work must happen in the cloud.

Why Are Developers Moving AI Work Off the Cloud?

The origin story of Osaurus reveals a fundamental frustration with cloud-based AI pricing. Terence Pae, a former software engineer at Tesla and Netflix, was building a desktop AI companion called Dinoki when customers asked a simple but pointed question: why should they pay for the app if they still had to pay for tokens, the usage units that cloud AI companies charge for processing prompts and generating responses?

"That's how Osaurus started," Pae explained. "You can do pretty much everything on your Mac locally, like browsing your files, accessing your browser, accessing your system configurations. I figured this would be a great way to position Osaurus as a personal AI for individuals."

Terence Pae, Co-founder of Osaurus

This insight sparked a broader question: what if users could run AI models directly on their own hardware instead of relying on expensive cloud subscriptions? Pae began building Osaurus in public as an open-source project, iterating on features and addressing bugs along the way.

How Does Osaurus Actually Work?

Osaurus functions as what's called a "harness," a control layer that connects different AI models, tools, and workflows through a single interface. The key innovation is flexibility: users can run AI models locally on their Mac, connect to cloud providers like OpenAI and Anthropic, or mix and match depending on their needs.

The platform supports a wide range of models and services, giving users genuine choice about where their AI processing happens:

• Local Models: MiniMax M2.5, Gemma 4, Qwen 3.6, GPT-OSS, Llama, DeepSeek V4, and Apple's own on-device foundation models
• Cloud Providers: OpenAI, Anthropic, Gemini, xAI/Grok, Venice AI, OpenRouter, Ollama, and LM Studio
• Native Integrations: Over 20 built-in plugins for Mail, Calendar, Vision, macOS utilities, spreadsheets, presentations, browser access, music, Git, filesystem operations, search, and data fetching

Unlike some competing tools that require technical expertise, Osaurus presents an easy-to-use interface designed for consumers rather than just developers. The platform also addresses security concerns by running AI operations in a hardware-isolated virtual sandbox, which limits the AI to a specific scope and keeps your computer and data protected.

What Are the Hardware Requirements for Running Local AI?

Running AI models locally is still resource-intensive and hardware-dependent. To run local models on Osaurus, your system needs at least 64 gigabytes of RAM. For larger models like DeepSeek V4, Pae recommends systems with approximately 128 gigabytes of RAM.

"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 stated.

Terence Pae, Co-founder of Osaurus

Pae believes these hardware requirements will decrease over time as local AI efficiency improves. The rapid pace of improvement in what researchers call "intelligence per wattage" suggests that running capable AI models on consumer hardware will become increasingly practical.

Pae

Steps to Get Started With Local AI on Your Mac

• Check Your Hardware: Ensure your Mac has at least 64GB of RAM for basic local models, or 128GB for larger models like DeepSeek V4
• Download Osaurus: The platform is available as a free, open-source download for Mac users who want to run AI locally
• Choose Your Model: Select from local models like Llama or Gemma, or connect to cloud providers like OpenAI and Anthropic depending on your needs
• Enable Plugins: Activate the native integrations you need, such as browser access, file management, or email integration
• Set Security Preferences: Configure the hardware-isolated sandbox to control what data and system access the AI can use

How Could Local AI Change Enterprise and Healthcare?

The founders of Osaurus, including co-founder Sam Yoo, are currently participating in the Alliance, a New York-based startup accelerator. They're exploring how local AI could serve businesses in regulated industries like legal services and healthcare, where running language models on-premises rather than in the cloud could address strict privacy requirements.

Pae believes that as local AI capabilities grow, the demand for massive cloud AI data centers could decline. Instead of relying on centralized cloud infrastructure, organizations could deploy Mac Studio computers on-premises, using substantially less power while maintaining the same AI capabilities.

"We're seeing this explosive growth in the AI space where cloud AI providers have to scale up using data centers and infrastructure, but we feel like people haven't really seen the value of the local AI yet. Instead of relying on the cloud, they can actually deploy a Mac Studio on-prem, and it should use substantially less power. You still have the capabilities of the cloud, but you will not be dependent on a data center to be able to run that AI," Pae noted.

Terence Pae, Co-founder of Osaurus

Osaurus competes with other tools that enable local AI, including Ollama, Msty, and LM Studio, but differentiates itself by offering a more user-friendly interface designed for non-technical consumers alongside its technical capabilities. With over 112,000 downloads since its launch, the platform is gaining traction as an alternative to cloud-only AI workflows.

The emergence of tools like Osaurus signals a broader shift in how developers and organizations think about AI infrastructure. Rather than accepting that all AI work must happen in the cloud, users now have practical options to keep their data, files, and AI processing on their own hardware while maintaining flexibility to use cloud services when it makes sense. As local AI models continue to improve in capability and efficiency, this hybrid approach may become the default rather than the exception.