# llama.cpp v b11527 adds 2D WebGPU workgroups and new binaries

> The open‑source LLM runtime updates its WebGPU engine to use 2‑D dispatch and publishes fresh builds for macOS, iOS and Linux.

Oossa · 2026-10-09 · https://oossa.com/en/llama-cpp-v-b11527-adds-2d-webgpu-workgroups-and-new-binaries

The ggml‑org team released llama.cpp version b11527 on 2026-10-09. The update switches all WebGPU operations to 2‑D workgroup dispatch, removing the older 1‑D method. It also bundles new pre‑compiled binaries for macOS (Apple Silicon and Intel), iOS, and several Ubuntu configurations, including Vulkan and CUDA builds. The changes are aimed at smoother GPU performance in browsers and easier setup on supported devices.

## The facts

- Release tag: b11527 (2026‑10‑09)
- WebGPU now uses 2‑D workgroup dispatch for all ops

## Why it matters

Developers can run LLMs faster in WebGPU‑enabled browsers without custom kernel code.

## Sources & references

1. [ggml-org/llama.cpp b11527](https://github.com/ggml-org/llama.cpp/releases/tag/b11527) – llama.cpp, 2026-10-09

Last updated: 2026-10-09