# llama.cpp release b11540 adds SYCL support for faster MoE decoding

> The open‑source llama.cpp library released version b11540 on Oct 10 2026, adding SYCL acceleration for MXFP4 MoE models and new binaries for macOS, iOS and Linux.

Oossa · 2026-10-10 · https://oossa.com/en/llama-cpp-release-b11540-adds-sycl-support-for-faster-moe-decoding

The ggml‑org team published llama.cpp version b11540 on Oct 10 2026. The update adds SYCL support to speed up MXFP4 mixture‑of‑experts (MoE) models using arithmetic decoding and weight reordering. It also ships pre‑built binaries for macOS Apple Silicon, Intel, iOS, and several Ubuntu variants (CPU, Vulkan and CUDA). The changes are listed under the tag b11540 on GitHub.

## The facts

- Release tag b11540 published Oct 10 2026
- SYCL acceleration added for MXFP4 MoE models

## Why it matters

Developers can now run larger MoE models faster on GPUs that support SYCL, expanding hardware options for llama.cpp users.

## Sources & references

1. [ggml-org/llama.cpp b11540](https://github.com/ggml-org/llama.cpp/releases/tag/b11540) – llama.cpp, 2026-10-10

Last updated: 2026-10-10