llama.cpp rolls up Aug 22 backend and model fixes

инструменты официальный 1 ист. ~1 мин

Between 2026-08-22 08:13 and 23:40 UTC, llama.cpp tagged ten commit-releases (b10577–b10588) on the master branch. Notable items: PAD_REFLECT_1D op added to the Vulkan backend, ROCm CI restored for Ubuntu, DSpark support for bailingmoe3, mtmd vision+audio support for dots3-note, common/json.h abstraction, concat op optimized via row-level memcpy, mtmd switched to ggml_rope_set_offset, draft-mtp with embeddings fixed.

Почему это важно

Rolling llama.cpp updates land Vulkan backend expansion and mtmd multimodal model support

Важность: 2/5

Standard release

llama-cpp inference ggml release vulkan

Источники

официальный Releases · ggml-org/llama.cpp