We Built a Local Model Arena in 30 Minutes — Infrastructure Mattered More Than the App

LLMesh is an open-source inference broker that turns any machines on your network into a shared LLM compute pool — one API endpoint, any backend, any machine. We built a Model Arena in 30 minutes to prove it: same app code works on a laptop, across a team, and in production without changing a line.