b10155

ggml-org/llama.cppb10155Jul 27, 2026by github-actions[bot]same day as b10154

What changed since b10154

medium risk

b10155 adds MiMo-V2.5 audio input support to mtmd via a new RVQ-based model implementation and GGUF converter.

New features

  • Multimodal (mtmd) support for MiMo-V2.5 audio input using a Residual Vector Quantization (RVQ) based model.
  • New GGUF converter for MiMo audio models, enabling conversion of the model weights for use with llama.cpp.

Improvements

  • C++ inference implementation added for the RVQ audio path, integrating it into the existing mtmd pipeline.
  • Cross-platform binary matrix (macOS, Linux CPU/Vulkan/ROCm/OpenVINO) carried forward unchanged for the new build.

Upgrade notes

No configuration or CLI changes are required; existing text and multimodal workflows are unaffected. To use MiMo-V2.5 audio, convert the model with the newly added GGUF converter before loading it through mtmd. This is a routine build bump, so pinned deployments can upgrade without migration steps.

Analyzed Nov 1, 2026

Upgrading?

Release notes summary

Adds support for the MiMo-V2.5 audio model, which utilizes Residual Vector Quantization (RVQ) for audio input processing.

Key Highlights

  • MiMo-V2.5 audio model support
  • RVQ-based model implementation

Release Notes (excerpt)

<details open> mtmd: support MiMo-V2.5 audio input (RVQ-based model) (#26190) * gguf converter for mimo audio * fix conv * cpp impl * nits * nits 2 </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)…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories