December 2025 monthly summary focusing on business value and technical achievements. Highlights include the delivery of a new count_equal tensor operation for Metal backends and improvements to memory management and performance. Work spans two repositories (ggml-org/ggml and ggml-org/llama.cpp) with a consistent implementation and cross-repo documentation updates. Key outcomes: - Enabled efficient counting of equal elements between tensors on Apple Metal, accelerating compute-intensive workloads and broadening deployment options on Metal-enabled devices. - Improved correctness and stability through memory initializations (zeroing dst buffers) and data-type adjustments (shmem to i32). - Code hygiene and maintenance enhancements, including removal of trailing whitespace, documentation table updates, and alignment with review feedback (e.g., doc updates, removal of outdated BLAS references in Metal docs). - Strong cross-team collaboration with co-authored contributions to ensure robust integration and future extensibility of tensor ops on Metal.