b9596

ggml-org/llama.cppb9596Jun 11, 2026by github-actions[bot]same day as b9594

What changed since b9594

low risk

b9596 adds router-mode log filtering on top of b9594's vocab normalizer refactor, with no API or behavior breaks.

New features

  • Server router mode now suppresses log lines that are irrelevant in that mode, reducing noise in multi-model deployments.

Improvements

  • Refreshed prebuilt binaries across macOS, Linux (CPU/Vulkan/ROCm/OpenVINO), Windows (CUDA 12/13, Vulkan, HIP), Android, and openEuler, plus an updated iOS XCFramework.

Upgrade notes

This is a two-build delta with only a server logging tweak and rebuilt artifacts, so a drop-in replacement should suffice. The vocab normalizer refactor and strip_accents option from b9594 remain available; no config or migration steps are required.

Analyzed Oct 31, 2026

Upgrading?

Release notes summary

Server log optimization for router mode and maintenance of various binary builds.

Key Highlights

  • Skip unused log lines in server router mode
  • Maintained builds for macOS (Apple Silicon, Intel), Linux (CPU, Vulkan, ROCm, OpenVINO), Windows (CPU, CUDA 12/13, Vulkan, HIP), Android, and openEuler
  • iOS XCFramework updated

Release Notes (excerpt)

<details open> server: skip unused log lines on router mode (#24463) </details> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - Ubuntu x64 (CPU) - Ubuntu arm64 (CPU) - Ubuntu s390x (CPU) - Ubuntu x64 (Vulkan) - Ubuntu arm64 (Vulkan) - Ubuntu x64 (ROCm 7.2) - Ubuntu x64 (OpenVINO) - Ubuntu x64 (SYCL FP32) DISABLED **Android:** - Android arm64 (CPU) **Windows:** - Windows x64 (CPU)…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories