# llama.cpp release b11672 adds new macOS and Linux binaries

> The open-source LLaMA inference library posted version b11672 on Oct 16, 2026, with fresh binaries for Apple Silicon, Intel macOS and several Ubuntu builds.

Oossa · 2026-10-16 · https://oossa.com/en/llama-cpp-release-b11672-adds-new-macos-and-linux-binaries

The ggml‑org team released llama.cpp version b11672. It ships pre‑compiled binaries for macOS Apple Silicon, macOS Intel, iOS, and multiple Ubuntu CPU, Vulkan and CUDA configurations. The release also notes a change to the mtmd lightonocr tool: it no longer forces the longest edge to be 1540 pixels, using the native image size by default.

## The facts

- Release date: Oct 16, 2026 ("Fri Oct 16 2026 04:22:54 GMT+0200").
- Binaries include macOS arm64, macOS x64, iOS XCFramework, Ubuntu x64/arm64/s390x (CPU, Vulkan, CUDA).

## Why it matters

Developers can now run LLaMA models on more hardware setups without building the code themselves.

## Sources & references

1. [ggml-org/llama.cpp b11672](https://github.com/ggml-org/llama.cpp/releases/tag/b11672) – llama.cpp, 2026-10-16

Last updated: 2026-10-16