logoalt Hacker News

UncleOxidantyesterday at 4:53 PM0 repliesview on HN

LMStudio, but it uses llama.cpp to run inference, so yeah. This is with the vulkan backend, not ROCm.