llama.cpp

56672bf6 - Merge commit 'refs/pull/27742/head' of github.com:ggerganov/llama.cpp into qwen4exp/qwen3.8-flash-next