It's here. The first fully native agent harness, built for Windows. No electron in sight. Osaurus for Windows is almost ready for beta. Get ready for a team of agents running right on your PC, local and private. Watch Qwen3.8-27B set up a team of agents to perform a... code review.show more

Osaurus for Windows is built with WinUI & C# on .NET 10 and runs on x64 and ARM64. It idles ~120MB, leaving your RAM for the rest of the work you're doing. Use a Local model, bring your own API key, or use your existing subscription. Come help shape the future of local AI on Windows. Drop your email here for access to the beta:

The macOS version is open source, MIT licensed:

That's not native bruh. Pure native is u use c++ or Delphi and that app will go down 10-20mb ram usage and <10mb binary often single portable binary. Still u could compile that to native aot. 100mb is a lot of memory. I would run at least 5 apps with that RAM space.

Qwen3.8-27B doing the whole review locally is nuts.

Native local execution changes the product boundary in a meaningful way. The interesting constraint is not just model size, but how the harness isolates agents while still giving them useful files and tools. How are you thinking about permissioning across a team of local agents?

You should probably get a DGX Spark or something and connect it to Osaurus for a better experience instead of using your work device

No electron in sight is the interesting part. Native harnesses avoid the whole Chromium memory tax, though you end up owning windowing and rendering yourself. Curious how you handle the browser side when agents need to click through real pages.

Haha.. I am just discovering this. That is the way. We are building the macOS version.

C# wtf native shit , go full native. That's a wrapper of a wrapper. Worst than java. Python-level shit.

will it use the NPU? 👀