llama.cpp b10470 improves release pipeline with explicit tag push

The llama.cpp project, maintained by ggml-org and the backbone of local LLM inference, has shipped version b10470. This release is a maintenance update that addresses a reliability issue in the project's release automation. Specifically, commit 34af94c (created on GitHub.com and signed with a verified GPG key) modifies release.yml to add an explicit "Create and push git tag" step before the "Create release" step. Instead of relying on the Releases API to create the tag as a side effect, the new workflow uses git tag and pushes it with the configured deploy key. The step is idempotent—if the tag already exists (e.g., on a re-run), creation and push are skipped.

While no new model features are introduced, b10470 includes prebuilt binaries for nearly every platform and backend the project supports. Users can download macOS Apple Silicon (including a KleidiAI-optimized variant), macOS Intel, iOS XCFramework, Linux x64/arm64/s390x with CPU, Vulkan, ROCm 7.14, OpenVINO, and SYCL builds, Android arm64, and Windows x64/arm64 with CUDA 12.4/13.3, Vulkan, OpenCL, and more. This wide distribution ensures that developers running local inference on edge devices, gaming GPUs, or enterprise hardware can stay on the latest stable release without needing to compile from source.