llama.cpp Release b10647 Introduces Video CLI Arguments
The b10647 release of llama.cpp introduces new `--video-*` command-line arguments and minor helper initialization updates, specifically adding `mtmd_helper_init_opt`. While this is a routine incremental release, the addition of video-related CLI arguments suggests ongoing work to improve support for multimodal models that process video inputs. The release includes pre-built binaries for various platforms, though the macOS Apple Silicon build with KleidiAI acceleration enabled remains disabled.
## BACKGROUND
llama.cpp is a highly popular open-source C/C++ LLM inference engine designed to run models like Meta's Llama locally with high performance. KleidiAI is an open-source library by Arm designed to accelerate AI workloads on Arm-based devices.