qwen3.8-27b-in-c

Run 27B-parameter Qwen3.8 locally on a laptop CPU with native C inference, no GPU or Python, at 2.52 token/s.

GitHub

Official source

Start locally with the CLI guide