llama.cpp folds --mmap, --no-mmap, --mlock and --direct-io into one --load-mode flag

Release b10441 (15 August 2026) migrated the deprecated --mmap / --no-mmap / --mlock / --direct-io options to a single --load-mode argument across scripts, examples and docs — a breaking change for existing command lines and service files. The same day's builds added Kimi-K3 support with hybrid attention, latent mixture-of-experts and MXFP4 quantisation handling (b10448), MiniMax-Text-01 and MiniMax-M1 (b10437), and Vulkan work for Intel Xe; b10451 on 16 August bounds-checks LoRA tensor data.

What it means Anyone pinning llama.cpp in a Dockerfile, systemd unit or run script has a flag rename to make before the next bump — silent, and it fails at start-up rather than at build.