b10269

ggml-org/llama.cppb10269Aug 4, 2026by github-actions[bot]same day as b10268

What changed since b10268

low risk

b10269 adds a single model-loading fix for the dflash wo_a reshape, on top of b10268's macOS CI fix.

Improvements

  • Model loading is more robust for dflash wo_a checkpoints thanks to a corrected reshape during load.

Bug fixes

  • Fixed an incorrect reshape of the dflash wo_a tensor when loading the model, which could cause load failures or malformed weights.

Upgrade notes

This is a one-commit increment over b10268; no config, CLI, or API changes are involved. Only users loading dflash wo_a models need this build, and everyone else can upgrade without any migration steps.

Analyzed Oct 31, 2026

Upgrading?

Release notes summary

Fixes a reshape issue for the dflash wo_a model during loading.

Key Highlights

  • Fix for dflash wo_a reshape on load
  • Model loading stability improvement

Release Notes (excerpt)

<details open> models : fix dflash wo_a reshape on load (#26577) </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 7.2) - Ubuntu x64 (OpenVINO) - Ubuntu x64 (SYCL FP32) - Ubuntu x64 (SYCL FP16) **Android:**…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories