b9711

ggml-org/llama.cppb9711Jun 19, 2026by github-actions[bot]same day as b9707

What changed since b9707

low risk

Refactored LLaVA-UHD overview image handling and corrected implementation limits.

Improvements

  • Refactored LLaVA-UHD overview image handling to standardize the processing pipeline.
  • Improved code stability with corrections to overview image processing.
  • Added a dedicated 'overview' field to mtmd_image_preproc_out structure.

Bug fixes

  • Corrected implementation limits and logic in overview image handling.
  • Fixed stability issues in Multimodal image preprocessing.

Upgrade notes

This release refactors internal processing for LLaVA-UHD overview images. Ensure your Multimodal models are tested, as the output structure for mtmd_image_preproc_out now includes a dedicated 'overview' field.

Analyzed Oct 20, 2026

Upgrading?

Release notes summary

Refactors the handling of overview images in the Multimodal LLaVA-UHD model to standardize the processing pipeline.

Key Highlights

  • Refactored LLaVA-UHD overview image handling
  • Added a dedicated 'overview' field for mtmd_image_preproc_out
  • Improved code stability with corrections

New Features

  • MTMD overview image handling refactor

Release Notes (excerpt)

<details open> mtmd: refactor llava-uhd overview image handling (always use ov_img_first) (#24769) * add dedicated "overview" for mtmd_image_preproc_out * corrections * correct (again) * nits * nits (2) </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) - Ubuntu x64 (Vulkan) - Ubuntu arm64 (Vulkan)…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories