# llama.cpp update adds BF16 precision support

> The open‑source LLaMA inference library now handles BF16 math, improving performance on several backends.

Oossa · 2026-11-08 · https://oossa.com/en/llama-cpp-update-adds-bf16-precision-support

The ggml‑org team released llama.cpp version b12202. It adds support for the BF16 (brain‑floating‑point 16‑bit) precision in key matrix multiplications. The change mirrors the reference implementation and is now accepted by CUDA, Vulkan, OpenCL, Hexagon and other backends. Tests for flash‑attention now run with BF16 alongside the existing F32 mode.

## The facts

- Version b12202 released on 2026-11-08
- Adds GGML_PREC_BF16 handling across multiple backends

## Why it matters

Developers can run LLaMA models faster on hardware that prefers BF16, such as newer GPUs.

## Sources & references

1. [ggml-org/llama.cpp b12202](https://github.com/ggml-org/llama.cpp/releases/tag/b12202) – llama.cpp, 2026-11-08

Last updated: 2026-11-08