Support Fanxin Sun

Hi, I’m Fanxin. I’m building YFCE, an independent open-source effort to make advanced AI run locally—efficiently, transparently, and on hardware people can actually own.

The projects span the full stack: lmz losslessly compresses and packages models; lmsluice moves them through the fastest measured route; vram explores training beyond physical GPU memory; cuda4AS is building native CUDA compatibility for Apple silicon; and YFCE’s model-native OS and organism research explore voice, vision, spatial understanding, and safe adaptive machines.

Some components ship today; others are carefully measured research in progress. Results, limitations, and failures are published openly rather than hidden behind headline claims.

Your support helps fund test hardware, Apple and GPU access, CI, model evaluation, and the cross-platform experiments needed to turn these ideas into reliable open-source software.