ggml-org/llama.cpp

Feature Request: support Qwen3-omni, Qwen3-TTS and MiniCPM-o

Open

#17634 opened on Dec 1, 2025

View on GitHub
 (13 comments) (27 reactions) (0 assignees)C++ (110,169 stars) (18,202 forks)batch import
Review Complexity : Highenhancementgood first issuehelp wantedmtmdtts

Description

Prerequisites

  • I am running the latest code. Mention the version if possible as well.
  • I carefully followed the README.md.
  • I searched using keywords relevant to my issue to make sure that I am creating a new issue that is not already open (or closed).
  • I reviewed the Discussions, and have a new and useful enhancement to share.

Feature Description

support qwen omni audio generator

Motivation

currently omni can work but it does not provide a whisper-like api for tts and it does not support asr.

Possible Implementation

https://github.com/ggml-org/llama.cpp/issues/16885 https://github.com/ggml-org/llama.cpp/issues/16186 https://github.com/ggml-org/llama.cpp/issues/11347

Contributor guide

Tech stack
cpp
Domain
backendmachine learning
Issue type
feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
4
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
over 1 week
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
active
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
mostly clear
Prerequisites
C++model architecture understandingaudio processingmultimodal ML
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
20
Research direction
The issue requests support for Qwen3 omni, Qwen3 TTS, and MiniCPM o in llama.cpp. It links to other issues (#16885, #16186, #11347) that may provide context. To implement, first understand the existing multimodal support in llama.cpp by examining files like `examples/` or `ggml *` components. Look at how whisper integration is done for a reference API. The implementation will require adding new model architectures and possibly audio processing code.