Osaurus Brings Local and Cloud AI Models to Your Mac
Osaurus, an open-source Mac application, has emerged to bridge the gap between local and cloud AI models, enabling users to choose their preferred AI while keeping data on their own hardware. Founded by Terence Pae, the app originated from a need to offer AI without continuous token costs, prioritizing user control and privacy. It supports numerous models and tools, offering a user-friendly interface and security through a virtual sandbox, aiming to shift AI reliance from data centers to local machines.
A new open-source Mac application called Osaurus is positioning itself as a pivotal software layer in the evolving AI landscape, enabling users to seamlessly integrate and switch between local and cloud-based AI models. Launched by co-founder Terence Pae, formerly of Tesla and Netflix, Osaurus uniquely keeps users' data, memory, and tools securely on their own hardware, offering an appealing blend of flexibility, privacy, and control for Apple users.
The genesis of Osaurus stemmed from an earlier project, Dinoki, an "AI-powered Clippy" desktop companion. Pae observed that Dinoki customers questioned the value of paying for an app if they still incurred token costs from cloud AI providers. This insight spurred a deeper exploration into the viability and benefits of running AI directly on a user's machine, leading to the public, open-source development of Osaurus.
Today, Osaurus functions as a versatile AI harness, connecting a diverse array of locally hosted AI models with prominent cloud providers such as OpenAI and Anthropic. This architecture empowers users to select the AI model best suited for specific tasks, leveraging the unique strengths of different AIs while maintaining all critical data and associated tools within their personal Mac environment. Unlike many existing AI harnesses that cater primarily to developers via command-line interfaces, Osaurus distinguishes itself with an intuitive, user-friendly interface designed for consumers.
Security is a paramount concern, and Osaurus addresses this by running operations within a hardware-isolated, virtual sandbox. This robust containment mechanism limits the AI's scope, thereby protecting the user's computer and sensitive data from potential vulnerabilities that might exist in less secure, developer-focused tools.
Running AI models locally remains a resource-intensive endeavor, currently requiring significant hardware. For optimal performance, users will need at least 64 GB of RAM, with larger models like DeepSeek v4 necessitating approximately 128 GB. Despite these demands, Pae expresses strong confidence in the future of local AI, highlighting the rapid improvements in "intelligence per wattage." He notes that local AI capabilities have advanced dramatically in the past year, evolving from basic sentence completion to performing complex tasks like tool execution, code writing, and browser interaction.
Osaurus supports a broad spectrum of AI models, including MiniMax M2.5, Gemma 4, Qwen3.6, GPT-OSS, Llama, and DeepSeek V4, alongside Apple's on-device foundation models and Liquid AI's LFM family. For cloud integration, it connects with services like OpenAI, Anthropic, Gemini, xAI/Grok, Venice AI, OpenRouter, Ollama, and LM Studio. As a full Model Context Protocol (MCP) server, it also allows compatible clients to access over 20 native plugins for applications such as Mail, Calendar, Vision, macOS Use, XLSX, PPTX, Browser, Music, Git, Filesystem, and Search, with recent updates adding voice capabilities.
Since its launch approximately a year ago, the Osaurus project has garnered over 112,000 downloads, signaling significant interest from the Mac community. Pae and co-founder Sam Yoo are currently participating in the Alliance startup accelerator and are exploring future avenues, including offering Osaurus to enterprise clients in privacy-sensitive sectors like legal and healthcare. They envision local LLM deployments reducing the growing reliance on massive, power-hungry AI data centers, proposing on-premise Mac Studio setups as a more energy-efficient and secure alternative for substantial AI capabilities.
FAQ
Q: What are the primary benefits of using Osaurus?
A: Osaurus allows users to combine local and cloud AI models, offering flexibility in choosing the best AI for a task while keeping all user data, memory, and tools on their Mac for enhanced privacy and control. It also aims to reduce dependence on cloud data centers.
Q: What are the hardware requirements for running local AI models with Osaurus?
A: To run local models, a Mac system will generally require at least 64 GB of RAM. For larger models like DeepSeek v4, approximately 128 GB of RAM is recommended.
Q: How does Osaurus address security concerns for local AI?
A: Osaurus runs its AI operations within a hardware-isolated, virtual sandbox. This limits the AI's access and scope, thereby safeguarding the user's computer and personal data.
Related articles
PlayStation Showcase Chat Swamped by Demands for Destiny 3
PlayStation's recent State of Play showcase was largely overshadowed by an impassioned fan campaign in the Twitch chat, demanding 'Destiny 3'. Amidst reveals for new PS5 games, the chat was relentlessly spammed with #WeWantDestiny3, fueled by the unexpected sunsetting of Destiny 2 and the reported absence of a direct sequel. This digital protest reflects widespread community frustration, amplified by a popular streamer and a petition with over 330,000 signatures.
Microsoft Unveils ASSERT, Simplifying AI Behavior Testing with Text
Microsoft has launched ASSERT, an open-source framework designed to simplify AI behavior testing. It enables developers to create comprehensive, application-specific evaluations using natural language descriptions, ensuring AI systems act as intended for particular products and services. The tool translates high-level goals into structured tests, generates scenarios, scores results, and logs execution paths.
Trump Orders Voluntary AI Model Review Before Release
President Trump has signed an executive order creating a voluntary framework for AI companies to share advanced models with the federal government before release. This initiative aims to bolster secure innovation and protect critical infrastructure, reflecting a shift from the administration's previous hands-off approach to AI safety. Companies opting for pre-release review may receive confidentiality protections.
Quick Share Meets AirDrop: A Welcome Cross-Platform Step
Quick Verdict: A Much-Anticipated Bridge For years, seamless file sharing between Android and iOS devices has been a frustrating chasm, often requiring clunky workarounds or third-party apps. This month, Google is
Blue Origin's New Glenn Explosion: Key Components Survive, 2026
Blue Origin announced that critical fuel tanks and key launch pad components survived last week's New Glenn rocket explosion, paving a faster path back to flight. CEO Dave Limp pledges a return to orbital missions before year-end, which is crucial for NASA's Artemis lunar program to maintain its tight schedule for crewed landings.
ZeroDrift raises $10M to protect AI models from themselves: AI
ZeroDrift, an AI compliance startup, has secured $10 million in seed funding from investors like a16z Speedrun. The company's service acts as a crucial intermediary, detecting compliance violations in AI-generated messages and rewriting them to meet regulatory standards like SOC 2 and GDPR. This rapid, oversubscribed funding round highlights the urgent demand for robust AI governance solutions as businesses scale AI adoption.