b10433

ggml-org/llama.cppb10433Aug 14, 2026by github-actions[bot]same day as b10431

What changed since b10431

low risk

b10433 is a ggml sync-only release with no functional delta from b10431 beyond code consolidation.

Improvements

  • GGML submodule synchronized to a newer upstream revision.
  • Internal code consolidation and general stability touch-ups from the sync.
  • No API or behavior changes are documented, so existing builds should behave identically.

Upgrade notes

This is effectively a housekeeping bump: no config, CLI, or API changes are described, so a rebuild and re-pin should suffice. If you depend on the SSM recurrent-state rollback work, it remains available as introduced in b10431. Verify your backend binaries still match your pinned CUDA/ROCm/Vulkan toolchain after the sync.

Analyzed Oct 24, 2026

Upgrading?

Release notes summary

A sync release that consolidates ggml changes without introducing new features or breaking changes.

Key Highlights

  • GGML synchronization
  • Code consolidation
  • Stability improvements

Release Notes (excerpt)

<details open> sync : ggml </details> **Website:** - <https://llama.app> **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 7.14)DISABLED - Ubuntu x64 (OpenVINO) - Ubuntu x64 (SYCL FP32) - Ubuntu x64 (SYCL FP16) **Android:** - Android arm64 (CPU) **Windows:**…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories