# llama.cpp update improves backend prefetch handling

> The b12328 release adds smarter prefetch logic and skips MoE cache on unsupported devices.

Oossa · 2026-11-12 · https://oossa.com/en/llama-cpp-update-improves-backend-prefetch-handling

The ggml‑org/llama.cpp project released version b12328 on 2026-11-12. The update changes the backend to prefetch only inputs that were computed by earlier splits, avoiding reads of tensors that will be overwritten later. It also adds a check to skip the MoE (mixture‑of‑experts) cache when the current device cannot run the cached operation, preventing unnecessary data copies.

## The facts

- Release version b12328 published on 2026-11-12
- Adds prefetch‑only‑earlier‑splits logic and MoE cache skip on unsupported devices

## Why it matters

Developers can expect lower memory traffic and faster inference on hardware that cannot handle MoE workloads.

## Sources & references

1. [ggml-org/llama.cpp b12328](https://github.com/ggml-org/llama.cpp/releases/tag/b12328) – llama.cpp, 2026-11-12

Last updated: 2026-11-12