b9871

ggml-org/llama.cppb9871Jul 4, 2026by github-actions[bot]1 days after b9870

What changed since b9870

low risk

b9871 fixes a broken CPU concat implementation for quantized tensor types and adds regression tests.

Improvements

  • CPU concatenation of quantized tensors now produces correct results instead of failing or returning garbage.
  • Added concat test coverage for quantized types to guard against future regressions.

Bug fixes

  • Repaired the CPU concat kernel that was broken for quantized tensor types.

Upgrade notes

This is a drop-in patch release with no config or API changes; upgrading is safe for anyone running quantized models on CPU, especially workloads that concatenate tensors. No migration steps are required.

Analyzed Oct 30, 2026

Upgrading?

Release notes summary

Fixes a broken CPU implementation for concatenating quantized tensors. The previous implementation likely failed or produced incorrect results for quantized types.

Key Highlights

  • Fixed broken CPU concat implementation for quantized types
  • Added tests for quantized types to prevent regression
  • Restores correct tensor concatenation behavior on CPU

New Features

  • Fixed CPU concat implementation for quantized tensors

Release Notes (excerpt)

<details open> ggml : fix broken CPU concat implementation for quantized types (#25247) * ggml : fix broken CPU concat implementation for quantized types * tests : concat tests for quantized types --------- Co-authored-by: Stanisław Szymczyk <sszymczy@gmail.com> </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)…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories