Ornith 1.5 35B Q5 runs locally inside GitHub Copilot on a Mac M3 Max

DanWahlin · x · 2026-08-21

Dan Wahlin wrapped up a video showing the new Ornith 1.5 35B Q5 quantized model running locally inside the GitHub Copilot app. Tested on a Mac M3 Max with 64GB RAM, it wasn't fast (as expected) but worked well; the video is coming soon. He also jokingly asked NVIDIA for a spare DGX Spark, calling it a "need" after his wife banned more gear spending.

Original post →

More from coding & agent

coding & agent channel →