The ggml‑org team published llama.cpp version b11797 on Oct 21, 2026. The update adds IM2COL_3D support to the ggml‑opencl backend, a change contributed by OpenCode and Qualcomm engineer Li He. New pre‑built binaries are provided for macOS (Apple Silicon and Intel), iOS, and several Ubuntu flavours, including Vulkan builds. The release also aligns kernel and host types in OpenCL and fixes compatibility with older Adreno GPUs.
Why it matters
Developers can now run more efficient 3‑D convolution operations on OpenCL GPUs, improving performance for AI models on compatible hardware.
