Osaurus is a free, open-source AI application that runs entirely on your Mac using Apple Silicon, ensuring your data never leaves your machine.
Released today (version 0.22.22), it lets you use local AI models for private conversations and file processing, with the option to add ChatGPT, Claude, or Gemini for specific tasks while maintaining a shared memory across all of them.
It works offline by design and collects no analytics, addressing privacy concerns common to cloud-based AI tools.
What happened
Osaurus, an MIT-licensed application, launched version 0.22.22 today. It runs open-source AI models directly on your Mac using Apple Silicon, keeping all data local. Users can add ChatGPT, Claude, or Gemini when needed while maintaining a shared memory across all models.
Why it matters
Most AI tools send user input to remote servers, creating privacy and data security concerns. Osaurus addresses this by keeping conversations, code, and files entirely on your machine—your data never leaves unless you explicitly choose to send it to a paid service. The app works offline and does not collect analytics, making it valuable for users handling sensitive information or working in environments without reliable connectivity.
What to watch
Osaurus requires Apple Silicon and macOS 15.5 or later. The software is open-source (MIT licensed), built in public, and available for download; users can read, fork, and ship the code with their own projects. The app supports autonomous agents with voice control, folder watchers, and browser plugins, not just chat.
Osaurus launched today with version 0.22.22, positioning itself as a privacy-first alternative to cloud-dependent AI tools. The application runs open-source AI models directly on Apple Silicon Macs, eliminating the need to send data to external servers. Users can integrate ChatGPT, Claude, or Gemini as optional add-ons for tasks that benefit from those services, while maintaining a shared memory context across all connected models—enabling a seamless workflow where local and cloud-based AI work together.
The core technical premise is straightforward: local models stay on your machine, so conversations, code, and files remain private by default. Osaurus works entirely offline (Wi-Fi off) and does not collect analytics, embodying a design philosophy the creators describe as "off the grid. By design." The application supports advanced use cases beyond simple chat, including autonomous agents with voice control, folder watchers that can monitor directories for tasks, browser plugins, and parallel job execution—effectively positioning it as a local AI assistant aware of the user's environment and capable of reading, writing, and executing actions within it.
The distribution model reinforces the privacy commitment. Osaurus is MIT licensed, built in public, and open to forking and redistribution. This approach invites scrutiny and allows developers to verify the code does what it claims—no hidden telemetry or unexpected cloud calls. System requirements are modest but specific: Apple Silicon (the M-series chips) and macOS 15.5 or later. The application is free to download, removing a financial barrier to adoption and signaling that the creators view this as a foundational tool rather than a premium service.
Osaurus addresses a growing concern in the AI landscape: the tension between convenience and privacy. While most AI tools funnel user input to remote servers for processing—a model that enables powerful services but raises data security and compliance questions—Osaurus inverts that dynamic by keeping models and computation local to the user's machine. This design philosophy is particularly relevant for users handling sensitive code, proprietary documents, or personal information, as well as those working in environments where cloud connectivity is unreliable or restricted.
The application's architecture allows users to opt into cloud services selectively. Rather than forcing an all-or-nothing choice between privacy and capability, Osaurus lets users run local open models (which are free and transparent) for routine tasks while maintaining the ability to call ChatGPT, Claude, or Gemini for specific jobs that might benefit from those services' scale or training. By sharing memory across all integrated models, the tool preserves context and continuity—a practical advantage over switching between separate tools. The MIT license and public-source commitment signal that the creators are betting on user trust and long-term adoption through openness, not vendor lock-in.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
Ask AI anything about this article. Q&As are published on this page for other readers too.
BofA Global Research examined employment across 206 US industries and found almost no correlation between an i…
Z.ai announced GLM-5.3, a model with ~750B parameters that has surpassed Moonshot AI's Kimi K3 and matched or…
Anthropic will offer a watermark detection API allowing third-party developers to identify text created by Cla…
A developer released Dictata v0.1.0, a Windows application that transcribes speech locally using Whisper (an A…
WeaveScope, a tracing and monitoring platform for Elixir-based AI agents, entered early access
An essay compares large language model (LLM) training to baking bread, specifically cold-proofed focaccia
The AI news that matters, in one minute each morning.
Sign up free