KoboldCpp v1.124 adds YuE2 music generation and dozens of TTS models
KoboldCpp 1.124 introduces a small subset of popular models from audio.cpp, expanding its built‑in Music Gen and TTS Gen capabilities. The release supports the YuE2 music generation model in three configurations – a lightweight version without an LLM, a small‑LLM option that adds a 4B language model for lyric creation (recommended), and a large‑LLM variant for high‑end GPUs.
Key points
- KoboldCpp 1.124 adds YuE2 music generation with optional 4B LLM for lyrics
- AI Horde now supports Chat Completions workers for jinja, tool calls, streaming and vision
The update also brings new image generation support via sd.cpp sync, adding models such as MOSS Image, Fibo 2 Edit, Flux 3 Klein, and updates to existing ones like Krea 2 Turbo and Qwen Image 2.1. AI Horde now supports Chat Completions workers, enabling jinja, reasoning, tool calls, streaming, and vision. Various performance and stability fixes round out the release, including CUDA‑graph enablement, parallel context handling, and improved SWA for Gemma4 models.
Users can launch the updated binary for Windows, Linux, macOS, or AMD‑Vulkan builds and access the MusicUI at http://localhost:5001/musicui, which now includes a transcribe tab for Whisper models and a directory of supported TTS models.
Model pages: MOSS-Image → · Mistral Large 4 →
Koboldcpp v1.124 released
github.com · 31 October 2026Loading the full article…
This text was published by github.com. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
1source- Reddit discussionreddit.com
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Agents & Tools
All →- Google expands Gemini and Search integrations to third-party apps · 2 src
- Google Gemini adds custom skills and external app integration · 1 src
- Anthropic study finds experts follow AI agent's hidden preference 86% of the time · 9 src
- Apple launches Apple Intelligence features in China with iOS 27.3 beta · 1 src
- Google adds social sharing to Gemini AI avatars · 1 src
Comments
via GitHub Discussions