{"id":14447,"date":"2026-11-17T11:03:10","date_gmt":"2026-11-17T02:03:10","guid":{"rendered":"https:\/\/localmodelwatch.tsuchitsuchi.com\/?p=14447"},"modified":"2026-11-17T11:12:34","modified_gmt":"2026-11-17T02:12:34","slug":"koboldcpp-1-124-released-2","status":"publish","type":"post","link":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/11\/17\/koboldcpp-1-124-released-2\/","title":{"rendered":"KoboldCpp 1.124 Released: YuE2, TTS, and Image Gen Upgrades"},"content":{"rendered":"<p><!-- lmw:facts --><\/p>\n<h2>At a Glance<\/h2>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th>Item<\/th>\n<th>Value<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Repository<\/td>\n<td><a href=\"https:\/\/github.com\/LostRuins\/koboldcpp\">LostRuins\/koboldcpp<\/a><\/td>\n<\/tr>\n<tr>\n<td>Version<\/td>\n<td><a href=\"https:\/\/github.com\/LostRuins\/koboldcpp\/releases\/tag\/v1.124\">v1.124<\/a><\/td>\n<\/tr>\n<tr>\n<td>Published<\/td>\n<td>2026-10-31<\/td>\n<\/tr>\n<tr>\n<td>License<\/td>\n<td>AGPL-3.0<\/td>\n<\/tr>\n<tr>\n<td>Source type<\/td>\n<td>Primary source (the publisher itself)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p><em>Values determined by this site&#8217;s code when the information was collected. Dates are JST.<\/em><\/p>\n<p><!-- \/lmw:facts --><\/p>\n<h2>Overview<\/h2>\n<p>KoboldCpp is a zero-installation, single-binary tool for running GGUF-format AI models locally via the KoboldAI UI. In this version 1.124, the integration of features from audio.cpp significantly enhances music generation (Music Gen) and text-to-speech (TTS Gen) capabilities.<\/p>\n<p>In particular, the addition of support for the long-awaited music generation model YuE2 is the biggest highlight of this update. This enriches the text-to-music generation experience in a local environment.<\/p>\n<h2>Key Changes<\/h2>\n<h3>Significant Expansion of Audio and Music Generation Features<\/h3>\n<p>A subset of audio.cpp has been integrated, greatly enhancing Music Gen and TTS Gen features. This enables multimodal generation in local environments.<\/p>\n<p>For music generation, the YuE2 model is now supported. Users can choose from templates such as &#8220;YuE2, no LLM&#8221; (a lightweight option providing only lyrics), &#8220;YuE2, small LLM&#8221; (recommended, with a built-in LLM for lyric generation), and &#8220;YuE2, big LLM&#8221; (for large GPUs). Additionally, MusicUI has been updated to support YuE2 and new TTS models. MusicUI (http:\/\/localhost:5001\/musicui) also adds a new &#8220;transcribe&#8221; tab, which is useful when reference transcripts are required for new TTS models.<\/p>\n<p>For text-to-speech (TTS Gen), numerous models have been added, including BreezeTTS 2, Ming 3.0 TTS, Qwen3-TTS (Base and VoiceDesign), VoxCPM2, Chatterbox, CosyVoice 3, Fish Audio S2 Pro, Kitten TTS 2, Maya1, NeuTTS 2E, OmniVoice, PocketTTS, Soprano 1.1 80M, Supertonic 3, and VibeVoice 1.5B. These include models capable of voice cloning and instruction-based control.<\/p>\n<h3>Addition of Image Generation Models<\/h3>\n<p>Through synchronization with sd.cpp, support for new image generation models has been added. Specifically, MOSS Image, Fibo 2 Edit, and Flux 3 Klein are now available. Previously added models such as Krea 2 Turbo and Qwen Image 2.1 remain available as well.<\/p>\n<h3>Chat Completions Support in AI Horde<\/h3>\n<p>AI Horde now supports Chat Completions workers, which is automatically enabled by updating to the latest KoboldCpp. This allows you to use features like jinja, reasoning, tool calls, streaming, and vision via AI Horde. Users can enable this by toggling chat completions mode at <a href=\"https:\/\/lite.koboldai.net\">https:\/\/lite.koboldai.net<\/a>.<\/p>\n<h3>Inference Performance and Parallel Processing Improvements<\/h3>\n<p>When using parallel requests (<code>--parallelrequests<\/code>), a new <code>--paralleltotalcontext<\/code> option has been added to set the total amount of shared KV context pool. When handling large numbers of batch requests, increasing this value prevents large requests from queuing up. Limited cache reuse is now also possible in batch mode.<\/p>\n<p>Furthermore, CUDA graphs are now enabled across all architectures, upstream <code>--lazy-mode<\/code> and <code>--moe-cache-mib<\/code> for MoE cache have been exposed, and SWA (Sliding Window Attention) processing for Gemma4 models has been improved, resulting in more stable quality during rewinds.<\/p>\n<h3>Agent Features and Other Fixes<\/h3>\n<p>A <code>\/mcp<\/code> wizard has been added for KoboldCpp Agent to configure custom MCP endpoints. A <code>\/paste<\/code> mode for handling multi-line text has also been introduced. Other changes include improvements to web search functionality and support for Animated WebP as the default preview for video generation.<\/p>\n<h2>Supported Models and Hardware<\/h2>\n<p>This update brings support for many new model architectures and quantization formats, with major expansions particularly in audio\/music generation and image generation.<\/p>\n<p>For music generation, GGUF conversion and inference for the YuE2 model are now possible. In addition, numerous TTS models are supported, including BreezeTTS 2, Ming 3.0 TTS, Qwen3-TTS (Base\/VoiceDesign), VoxCPM2, Chatterbox, CosyVoice 3, Fish Audio S2 Pro, Kitten TTS 2, Maya1, NeuTTS 2E, OmniVoice, PocketTTS, Soprano 1.1 80M, Supertonic 3, and VibeVoice 1.5B.<\/p>\n<p>For image generation, synchronization with sd.cpp enables the use of MOSS Image, Fibo 2 Edit, and Flux 3 Klein, alongside previously added models like Krea 2 Turbo and Qwen Image 2.1.<\/p>\n<p>Regarding text models, support for latest models such as Nemotron 3.5 Bolt 74BA9B, Mistral Large 4, and LFM3 is also included.<\/p>\n<h2>How to Get It<\/h2>\n<p>Windows users should download and run <code>koboldcpp.exe<\/code> from the release. If you are not using an NVIDIA GPU or do not need CUDA, you can use the smaller <code>nocuda<\/code> version. If it fails to run on older CPUs or older NVIDIA GPUs, try the <code>oldpc<\/code> version (Cuda11 + AVX1).<\/p>\n<p>Linux users should use <code>koboldcpp-linux-x64<\/code>. If using an AMD GPU, it is recommended to try the Vulkan option of the <code>nocuda<\/code> build first. Rolling ROCm binaries are also available for Linux environments.<\/p>\n<p>macOS (Apple Silicon) users should use the <code>koboldcpp-mac-arm64<\/code> binary.<\/p>\n<p>For manual setup in an Android (Termux) environment, you can build it with the following steps:<\/p>\n<pre><code class=\"language-bash\">pkg update\npkg upgrade\npkg install openssl wget git python clang make\ngit clone https:\/\/github.com\/LostRuins\/koboldcpp.git\ncd koboldcpp\nmake\npython koboldcpp.py --model \/path\/to\/model.gguf\n<\/code><\/pre>\n<p><!-- lmw:releases --><\/p>\n<h2>Releases Since Our Last Article<\/h2>\n<p><em>Compiled by Local Model Watch from the project&#8217;s GitHub releases: the versions between this release and the last one we covered, which did not get separate articles.<\/em> <em>Full history: <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/engine-koboldcpp-en\/\">release tracker<\/a>.<\/em><\/p>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th>Version<\/th>\n<th>Released<\/th>\n<th>Release notes<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>v1.124.1<\/td>\n<td>2026-10-31<\/td>\n<td><a href=\"https:\/\/github.com\/LostRuins\/koboldcpp\/releases\/tag\/v1.124.1\">GitHub<\/a><\/td>\n<\/tr>\n<tr>\n<td>v1.124.2<\/td>\n<td>2026-10-31<\/td>\n<td><a href=\"https:\/\/github.com\/LostRuins\/koboldcpp\/releases\/tag\/v1.124.2\">GitHub<\/a><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p><!-- \/lmw:releases --><\/p>\n<p><!-- lmw:related --><\/p>\n<h2>Related Articles<\/h2>\n<ul>\n<li><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/11\/17\/koboldcpp-v1123-released\/\">KoboldCpp v1.123 Released: Major Agent Feature Upgrades<\/a><\/li>\n<li><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/09\/26\/koboldcpp-v1-122-released\/\">KoboldCpp v1.122 Released with Built-in Agentic Framework<\/a><\/li>\n<\/ul>\n<p><!-- \/lmw:related --><\/p>\n<p><!-- lmw:next-steps --><\/p>\n<h2>What to Read Next<\/h2>\n<ul>\n<li><strong>Follow this tool<\/strong> \u2192 <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/engine-koboldcpp-en\/\">KoboldCpp overview and release history (137 releases tracked)<\/a><\/li>\n<li><strong>Other inference engines and runtimes<\/strong> \u2192 <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/engine-llama-cpp-en\/\">llama.cpp<\/a> \/ <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/engine-ollama-en\/\">Ollama<\/a> \/ <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/engine-vllm-en\/\">vLLM<\/a><\/li>\n<\/ul>\n<p><!-- \/lmw:next-steps --><\/p>\n<h2>Sources<\/h2>\n<ul>\n<li><a href=\"https:\/\/github.com\/LostRuins\/koboldcpp\/releases\/tag\/v1.124\">LostRuins\/koboldcpp koboldcpp-1.124<\/a><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>KoboldCpp v1.124 is out, bringing audio.cpp integration, YuE2 music generation, numerous TTS models, image generation updates, and parallel request fixes.<\/p>\n","protected":false},"author":1,"featured_media":14446,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[105],"tags":[4482,4484,163,1277,1547,1144,213],"class_list":["post-14447","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-engines-and-tools","tag-ai-horde-en","tag-audio-gen-en","tag-gguf-en","tag-koboldcpp-en","tag-verified","tag-yue2-en","tag--en"],"lang":"en","translations":{"en":14447,"ja":14445},"_links":{"self":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts\/14447","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/comments?post=14447"}],"version-history":[{"count":2,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts\/14447\/revisions"}],"predecessor-version":[{"id":18016,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts\/14447\/revisions\/18016"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/media\/14446"}],"wp:attachment":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/media?parent=14447"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/categories?post=14447"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/tags?post=14447"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}