# llama.cpp b12247 release fixes Vulkan bugs on macOS

> The open‑source LLaMA inference library updates to version b12247, correcting out‑of‑bounds memory errors in MoltenVK for macOS and iOS.

Oossa · 2026-11-10 · https://oossa.com/en/llama-cpp-b12247-release-fixes-vulkan-bugs-on-macos

ggml‑org released llama.cpp version b12247 on November 10, 2026. The update fixes out‑of‑bounds reads and writes when using Vulkan through MoltenVK on macOS and iOS. It also resolves a sentinel‑mismatch error in the test‑backend‑ops suite and disables robustness checks for several math pipelines. Pre‑built binaries for macOS Apple Silicon, Intel, iOS, and various Ubuntu CPU targets are provided.

## The facts

- Version b12247 released on 2026-11-10
- Fixes out‑of‑bounds memory errors on MoltenVK (Vulkan on macOS/iOS)

## Why it matters

Mac developers can run LLaMA models on Apple devices via Vulkan without the crashes that the previous version caused.

## Sources & references

1. [ggml-org/llama.cpp b12247](https://github.com/ggml-org/llama.cpp/releases/tag/b12247) – llama.cpp, 2026-11-10

Last updated: 2026-11-10