b10251

ggml-org/llama.cppb10251Aug 4, 2026by github-actions[bot]same day as b10250

What changed since b10250

low risk

b10251 adds MTP (multimodal/multi-token prediction) support for GLM-4.7-Flash, on top of b10250's test-only changes.

New features

  • GLM-4.7-Flash now supports MTP processing, enabling its multimodal/multi-token prediction path in llama.cpp.

Improvements

  • Model resolution test coverage from b10250 (synthetic repo listings, loopback HTTP server, DSpark sidecar) remains in place and is unaffected by this release.

Upgrade notes

This is a single-commit delta over b10250 that only adds GLM-4.7-Flash MTP support; no config, CLI, or API changes are documented. Users not running GLM-4.7-Flash will see no behavioral difference, and no migration steps are required. Prebuilt binaries are published for the usual macOS, Linux (CPU/Vulkan/ROCm/OpenVINO) targets.

Analyzed Nov 1, 2026

Upgrading?

Release notes summary

Adds support for Multimodal (MTP) processing in the GLM-4.7-Flash model.

Key Highlights

  • GLM-4.7-Flash MTP support
  • Enhanced multimodal capabilities

Release Notes (excerpt)

<details open> model : support MTP in GLM-4.7-Flash (#24868) </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