b10245

ggml-org/llama.cppb10245Aug 3, 2026by github-actions[bot]same day as b10244

What changed since b10244

low risk

b10245 is a single-commit follow-up that fixes unused input tensors in the Minimax M3 graph.

Bug fixes

  • Corrected the Minimax M3 graph so input tensors that were previously ignored are now actually consumed during graph construction.
  • Added graph validation alongside the fix, catching malformed M3 graphs earlier.

Upgrade notes

This is a one-commit delta on top of the MSA memory refactor in b10244, with no API, CLI, or config changes. Users running Minimax M3 models should upgrade to get correct graph execution; others can treat it as routine housekeeping. No migration or rebuild steps beyond swapping the binaries.

Analyzed Nov 1, 2026

Upgrading?

Release notes summary

Fixed a bug in the Minimax M3 graph where certain input tensors were not being utilized.

Key Highlights

  • Fixed unused input tensors in the Minimax M3 graph
  • Graph validation and bug fix

New Features

  • Graph bug fix for Minimax M3

Release Notes (excerpt)

<details open> graph : fix unused input tensors in minimax m3 graph (#26519) </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