{"id":19124,"date":"2026-11-18T11:08:53","date_gmt":"2026-11-18T02:08:53","guid":{"rendered":"https:\/\/localmodelwatch.tsuchitsuchi.com\/2026\/11\/18\/kokoro-82m-gguf-released\/"},"modified":"2026-11-18T11:08:53","modified_gmt":"2026-11-18T02:08:53","slug":"kokoro-82m-gguf-released","status":"publish","type":"post","link":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/11\/18\/kokoro-82m-gguf-released\/","title":{"rendered":"Kokoro-82M-GGUF Lightweight Text-to-Speech Model: File List"},"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:\/\/huggingface.co\/ggml-org\/Kokoro-82M-GGUF\">ggml-org\/Kokoro-82M-GGUF<\/a><\/td>\n<\/tr>\n<tr>\n<td>Published<\/td>\n<td>2026-11-18<\/td>\n<\/tr>\n<tr>\n<td>License<\/td>\n<td>apache-2.0<\/td>\n<\/tr>\n<tr>\n<td>Formats<\/td>\n<td><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/format-gguf-en\/\">GGUF<\/a><\/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><code>ggml-org\/Kokoro-82M-GGUF<\/code> is a GGUF format conversion of the text-to-speech (TTS) model <code>hexgrad\/Kokoro-82M<\/code>. Published by ggml-org, the model card states that it can be used via the <code>llama-tts<\/code> command or the <code>\/tts<\/code> endpoint in <code>llama serve<\/code>.<\/p>\n<p>This repository is an automated conversion (noted on the card as an automated conversion via <a href=\"https:\/\/github.com\/ggml-org\/convert\">https:\/\/github.com\/ggml-org\/convert<\/a>). The following qualities and characteristics are based on the original model card for <code>hexgrad\/Kokoro-82M<\/code>.<\/p>\n<h2>Specifications<\/h2>\n<ul>\n<li>Original Model: <code>hexgrad\/Kokoro-82M<\/code> (an open-weight TTS model with 82 million parameters, according to the original model card)<\/li>\n<li>Distribution Format: GGUF<\/li>\n<li>Supported Languages and Voices: The Releases table in the original model card lists v1.0 (released January 27, 2025) as &#8220;8 &amp; 54&#8221; (8 languages, 54 voices). A voice list is stated to be in VOICES.md of the original model<\/li>\n<li>Voice Selection: Use <code>--tts-voice<\/code> to select one of the voices bundled with the model. The first character of the voice name indicates the language<\/li>\n<li><code>a<\/code>: American English &#8211; <code>b<\/code>: British English &#8211; <code>e<\/code>: Spanish &#8211; <code>f<\/code>: French &#8211; <code>h<\/code>: Hindi &#8211; <code>i<\/code>: Italian &#8211; <code>p<\/code>: Brazilian Portuguese &#8211; <code>j<\/code>: Japanese &#8211; <code>z<\/code>: Mandarin Chinese<\/li>\n<li>Input Constraints: Text input is in English only. For non-English languages, prompts must be converted into phonemes beforehand (e.g., output from espeak-ng), according to the documentation<\/li>\n<li>Training Data (from the original model card): Hundreds of hours of audio. Training costs are reported to be about $1000 for 1000 hours of training on an A100 80GB<\/li>\n<li>License Terms: Apache-2.0. The original model card explains that thanks to &#8220;Apache-licensed weights,&#8221; it can be deployed anywhere from production environments to personal projects, and states that it is used in numerous projects including commercial APIs<\/li>\n<\/ul>\n<p>Sampling rates, recommended step counts, and upper limits for output length are not specified in the documentation.<\/p>\n<h2>Performance and Quality<\/h2>\n<p>There are no benchmark figures or comparison tables with other models in the documentation. The original model card also lacks such a table, and evaluations are said to be located in a separate file, EVAL.md.<\/p>\n<p>The original model card states regarding Kokoro that it &#8220;achieves comparable quality to larger models while being significantly faster and more cost-efficient&#8221; despite its lightweight architecture. This is a claim by the publishers themselves, and third-party verification results are not included in the documentation.<\/p>\n<p>Sample outputs are said to be published in SAMPLES.md of the original model and via a demo (<a href=\"https:\/\/hf.co\/spaces\/hexgrad\/Kokoro-TTS\">https:\/\/hf.co\/spaces\/hexgrad\/Kokoro-TTS<\/a>).<\/p>\n<h2>Strengths and Use Cases<\/h2>\n<p>The card for the original model <code>hexgrad\/Kokoro-82M<\/code> introduces Kokoro as a lightweight, fast, and cost-efficient TTS model. With Apache-2.0 weights, it is stated to be deployable in environments ranging from production to personal projects. It is considered suitable for use cases where speech synthesis needs to be run locally as a resident service, or for integrating text-to-speech processing into local servers.<\/p>\n<p>In terms of use cases and languages, the following points can be gathered from the documentation:<\/p>\n<ul>\n<li>Multilingual Read-Aloud: Voices can be selected via prefixes for American English, British English, Spanish, French, Hindi, Italian, Brazilian Portuguese, Japanese, and Mandarin Chinese<\/li>\n<li>Direct English Text Processing: Non-English languages require prompts to be converted to phonemes beforehand (espeak-ng is cited as an example). Using languages like Japanese requires this preprocessing<\/li>\n<li>Server Usage: The <code>\/tts<\/code> endpoint of <code>llama serve<\/code> can be used<\/li>\n<\/ul>\n<p>This GGUF version is an automated conversion by ggml-org and does not include any new model tuning. The only change from the original model is the format adaptation to run with llama.cpp-based tools. Voice cloning capabilities are not mentioned in the documentation.<\/p>\n<p><!-- lmw:hardware --><\/p>\n<h2>Hardware Requirements<\/h2>\n<p><strong>Estimated requirements (calculated by Local Model Watch)<\/strong><\/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>Your VRAM<\/th>\n<th>Quantization<\/th>\n<th>File size<\/th>\n<th>Est. memory needed<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>4GB (laptop iGPU \/ phone class)<\/td>\n<td>F32<\/td>\n<td>0.1GB<\/td>\n<td>0.1GB<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p><em>Memory estimates add a 20% runtime overhead (KV cache, etc.) to the actual size of the distributed files. Actual usage varies with context length, batch size and inference engine. These figures are computed by this site from file sizes, not published by the model&#8217;s authors. The listed files are distributed in F32 (32-bit). Loading in a lower precision (BF16 or quantized) needs less memory, so treat this as the size of the files as distributed, not a minimum. This site has not measured the lower-precision figures. Compare with other models in our <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/vram-guide-en\/\">VRAM quick reference<\/a>. What the quantization names mean: <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/glossary-quantization-en\/\">glossary<\/a>.<\/em><\/p>\n<p><!-- \/lmw:hardware --><\/p>\n<p><!-- lmw:runnability --><\/p>\n<h2>Can You Run It Locally?<\/h2>\n<p>The publisher distributes this model as GGUF.<\/p>\n<p><strong>License \u2014 <code>apache-2.0<\/code> (Commercial use allowed):<\/strong> Permits commercial use, modification and redistribution. Redistribution requires including the license and stating changes; includes a patent grant.<\/p>\n<p><em>Compiled by this site&#8217;s code from the published formats and the license field. License summaries are not legal advice \u2014 check the publisher&#8217;s original terms before relying on them.<\/em><\/p>\n<p><!-- \/lmw:runnability --><\/p>\n<p><!-- lmw:files --><\/p>\n<h2>Distributed Files<\/h2>\n<p><em>Weight files published in <a href=\"https:\/\/huggingface.co\/ggml-org\/Kokoro-82M-GGUF\/tree\/main\">ggml-org\/Kokoro-82M-GGUF<\/a>, listed by this site from the Hugging Face API. Sizes are the actual file sizes.<\/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>File<\/th>\n<th>Size<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><code>Kokoro-82M-F32.gguf<\/code><\/td>\n<td>62MB<\/td>\n<\/tr>\n<tr>\n<td><code>mmproj-Kokoro-82M-F16.gguf<\/code><\/td>\n<td>208MB<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p><!-- \/lmw:files --><\/p>\n<h2>How to Get It<\/h2>\n<p>The distribution format is GGUF. <code>model_facts.gated<\/code> is false, and it can be obtained without agreeing to license terms. The repository is <code>ggml-org\/Kokoro-82M-GGUF<\/code>.<\/p>\n<p>To generate audio from the terminal, the card shows the following command (specifying the repository with <code>-hf<\/code>):<\/p>\n<pre><code class=\"language-bash\">llama-tts -hf ggml-org\/Kokoro-82M-GGUF -p &quot;Hello world&quot; --tts-voice af_heart --output out.wav\n<\/code><\/pre>\n<p>To use it as a server, start it with the following command and call the <code>\/tts<\/code> endpoint:<\/p>\n<pre><code class=\"language-bash\">llama serve -hf ggml-org\/Kokoro-82M-GGUF\n<\/code><\/pre>\n<p>Specify the voice name bundled with the model (e.g., <code>af_heart<\/code> in the example above) for <code>--tts-voice<\/code>. Detailed usage instructions can be found in the llama.cpp Kokoro documentation referenced by the card (sources below). There is no documentation regarding support for ComfyUI or Diffusers.<\/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\/ggml-org-releases-tinylfm3-omni-for-testing-gguf\/\">ggml-org Releases tinyLFM3-Omni-for-testing-gguf for CI<\/a><\/li>\n<li><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/11\/17\/minimax-music3-gguf-released\/\">MiniMax-Music3-GGUF Full-Length Music Generation Model: 12GB+ VRAM<\/a><\/li>\n<li><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/11\/17\/higgs-audio-v3-stt-gguf\/\">higgs-audio-v3-stt-GGUF Speech-to-Text Model: 4GB+ VRAM, File List<\/a><\/li>\n<li><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/11\/17\/soprano-1-1-80m-gguf-released\/\">Soprano-1.1-80M-GGUF Lightweight English TTS Model: File List<\/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>Find models by VRAM<\/strong> (This model runs from the 4GB tier) \u2192 <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/vram-guide-8gb-en\/\">Other models that run on a 8GB GPU<\/a><\/li>\n<li><strong>Formats this model is available in<\/strong> \u2192 <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/format-gguf-en\/\">GGUF format guide and models<\/a><\/li>\n<li><strong>Other models for the same task<\/strong> \u2192 <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/models-by-task-en\/#task-audio\">Other audio models<\/a><\/li>\n<\/ul>\n<p><!-- \/lmw:next-steps --><\/p>\n<h2>Sources<\/h2>\n<ul>\n<li><a href=\"https:\/\/huggingface.co\/ggml-org\/Kokoro-82M-GGUF\">https:\/\/huggingface.co\/ggml-org\/Kokoro-82M-GGUF<\/a><\/li>\n<li><a href=\"https:\/\/huggingface.co\/hexgrad\/Kokoro-82M\">https:\/\/huggingface.co\/hexgrad\/Kokoro-82M<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/ggml-org\/llama.cpp\/blob\/master\/docs\/multimodal\/tts.md#kokoro\">https:\/\/github.com\/ggml-org\/llama.cpp\/blob\/master\/docs\/multimodal\/tts.md#kokoro<\/a><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>Explore ggml-org\/Kokoro-82M-GGUF, a GGUF version of the Kokoro-82M text-to-speech model for running locally using llama.cpp tools.<\/p>\n","protected":false},"author":1,"featured_media":19123,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[203],"tags":[161,1970,163,5285,1073,5287,1547,213],"class_list":["post-19124","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-image-video-and-audio","tag-ggml-en","tag-ggml-org-en","tag-gguf-en","tag-kokoro-82m-gguf-en","tag-llama-cpp-en","tag-styletts2-ljspeech-en","tag-verified","tag--en"],"lang":"en","translations":{"en":19124,"ja":19122},"_links":{"self":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts\/19124","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=19124"}],"version-history":[{"count":0,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts\/19124\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/media\/19123"}],"wp:attachment":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/media?parent=19124"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/categories?post=19124"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/tags?post=19124"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}