{"id":5002,"date":"2026-09-26T23:32:26","date_gmt":"2026-09-26T14:32:26","guid":{"rendered":"https:\/\/localmodelwatch.tsuchitsuchi.com\/publisher-liquidai-en\/"},"modified":"2026-09-27T13:17:07","modified_gmt":"2026-09-27T04:17:07","slug":"publisher-liquidai-en","status":"publish","type":"page","link":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/publisher-liquidai-en\/","title":{"rendered":"Liquid AI Open Models: Model List, Licenses and Articles"},"content":{"rendered":"<h2>About Liquid AI<\/h2>\n<p><strong>Liquid AI<\/strong> is an AI company that publishes <strong>LFM (Liquid Foundation Models)<\/strong>, a family of small models designed to run on devices, through its <a href=\"https:\/\/huggingface.co\/LiquidAI\">LiquidAI organization on Hugging Face<\/a>. It has 176 models and updates frequently.<\/p>\n<p>LFM stands out for its <strong>hybrid architecture<\/strong>, not a plain Transformer, with an emphasis on small size and speed. The LFM2.5 model card describes the line as &#8220;a family of hybrid models designed for on-device deployment&#8221;.<\/p>\n<h2>What It Releases<\/h2>\n<ul>\n<li><strong>LFM2.5 (text)<\/strong>: 230M, 350M, 1.2B and 2.6B, plus the 8B-A1B MoE and others. According to the 2.6B model card, it has a 128K-token context and agentic post-training, and is competitive with models 4x larger on tool use and instruction following. It runs at 220 tokens\/s on an Apple M5 Max and 113 tokens\/s on an AMD Ryzen CPU in under 2.5 GB of memory.<\/li>\n<li><strong>LFM2.5-VL (models that also read images)<\/strong>: 450M to 3B, including Extract versions that pull information out of documents (we have a <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/model-liquidai-lfm2-5-vl-en\/\">family page<\/a>).<\/li>\n<li><strong>Japanese versions<\/strong>: models for Japanese such as LFM2.5-1.2B-JP and the audio model LFM2.5-Audio-1.5B-JP.<\/li>\n<li><strong>Helper models (DSpark)<\/strong>: models that speed up inference through speculative decoding, released for the main models.<\/li>\n<\/ul>\n<h2>Licenses<\/h2>\n<p><strong>Most LFM models use Liquid AI&#8217;s own &#8220;LFM Open License v1.0&#8221; (<code>lfm1.0<\/code>).<\/strong> It permits use, modification and redistribution in a form close to Apache-2.0, but <strong>commercial use is limited to users (legal entities) with annual revenue below US$10 million<\/strong>. Commercial use by an entity at or above that threshold is not licensed under it (the cap does not apply to non-commercial or research use by qualified non-profit organizations). This summary is not legal advice; read the license text before use.<\/p>\n<h2>Running Them Locally<\/h2>\n<ul>\n<li><strong>Official GGUF, MLX and ONNX builds are very plentiful.<\/strong> For almost all main models, Liquid AI itself publishes GGUF, MLX for Apple Silicon (4-bit to 8-bit and bf16) and ONNX builds, which run directly in llama.cpp, Ollama and LM Studio.<\/li>\n<li><strong>They are small enough to try on CPU-only computers and phones.<\/strong> A realistic approach is to start with a model around 1B and move up to the 2.6B or 8B-A1B if needed.<\/li>\n<\/ul>\n<p><em>Sources: the model card and license text of <a href=\"https:\/\/huggingface.co\/LiquidAI\/LFM2.5-2.6B\">LiquidAI\/LFM2.5-2.6B<\/a>; the model list of the LiquidAI organization on Hugging Face (all as of 2026-09-26).<\/em><\/p>\n<h2>Our Coverage and Data<\/h2>\n<p>Local Model Watch has published 1 article(s) on Liquid AI&#8217;s own models, 0 on third-party fine-tunes and quantized builds based on them, and 0 from its official blog. The model list below is collected by code from Hugging Face every day. Part of our <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/publishers-en\/\">publisher index<\/a>; model families link to their <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/models-en\/\">family pages<\/a>.<\/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>Hugging Face organization<\/td>\n<td><a href=\"https:\/\/huggingface.co\/LiquidAI\">LiquidAI<\/a><\/td>\n<\/tr>\n<tr>\n<td>Models on Hugging Face<\/td>\n<td>176<\/td>\n<\/tr>\n<tr>\n<td>Articles on its own models<\/td>\n<td>1<\/td>\n<\/tr>\n<tr>\n<td>Articles on derived models<\/td>\n<td>0<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<h2>Recent Models on Hugging Face<\/h2>\n<p>The publisher lists <strong>176 models<\/strong> in total on Hugging Face. Below are up to 15 of the most recent generative models (text, image, video, audio and so on), the scope of this site, per organization. Classifiers, feature extractors, research components and repositories without a model card are omitted, and the publisher&#8217;s own quantized builds and helper variants are folded into the row of the original model. Parameter counts are left blank for repositories that pack weights into integer types, where Hugging Face&#8217;s count is not the real parameter count.<\/p>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<p class=\"lmw-table-hint\" style=\"margin:0 0 4px;font-size:0.85em;opacity:0.7;\">\u2192 Scroll horizontally to see all columns<\/p>\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th>Released<\/th>\n<th>Model<\/th>\n<th>Task<\/th>\n<th>Parameters<\/th>\n<th>License<\/th>\n<th>Official variants<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>2026-08-12<\/td>\n<td><a href=\"https:\/\/huggingface.co\/LiquidAI\/LFM2.5-VL-3B\">LiquidAI\/LFM2.5-VL-3B<\/a><\/td>\n<td>vision-language<\/td>\n<td>3.1B<\/td>\n<td><code>lfm1.0<\/code><\/td>\n<td>DSpark-GGUF, DSpark, GGUF, MLX-bf16 and 5 more<\/td>\n<\/tr>\n<tr>\n<td>2026-07-29<\/td>\n<td><a href=\"https:\/\/huggingface.co\/LiquidAI\/LFM2.5-2.6B\">LiquidAI\/LFM2.5-2.6B<\/a><\/td>\n<td>text generation<\/td>\n<td>2.7B<\/td>\n<td><code>lfm1.0<\/code><\/td>\n<td>DSpark-GGUF, DSpark, MLX-nvfp4, MLX-mxfp4 and 10 more<\/td>\n<\/tr>\n<tr>\n<td>2026-07-11<\/td>\n<td><a href=\"https:\/\/huggingface.co\/LiquidAI\/LFM2.5-Encoder-350M-Diffusion\">LiquidAI\/LFM2.5-Encoder-350M-Diffusion<\/a><\/td>\n<td>text generation<\/td>\n<td>354M<\/td>\n<td><code>lfm1.0<\/code><\/td>\n<td>\u2014<\/td>\n<\/tr>\n<tr>\n<td>2026-06-25<\/td>\n<td><a href=\"https:\/\/huggingface.co\/LiquidAI\/LFM2.5-230M\">LiquidAI\/LFM2.5-230M<\/a><\/td>\n<td>text generation<\/td>\n<td>230M<\/td>\n<td><code>lfm1.0<\/code><\/td>\n<td>Base, ONNX, MLX-4bit, MLX-5bit and 4 more<\/td>\n<\/tr>\n<tr>\n<td>2026-05-28<\/td>\n<td><a href=\"https:\/\/huggingface.co\/LiquidAI\/LFM2.5-8B-A1B\">LiquidAI\/LFM2.5-8B-A1B<\/a><\/td>\n<td>text generation<\/td>\n<td>8.5B<\/td>\n<td><code>lfm1.0<\/code><\/td>\n<td>DSpark-GGUF, DSpark, Base, MLX-8bit and 6 more<\/td>\n<\/tr>\n<tr>\n<td>2026-05-26<\/td>\n<td><a href=\"https:\/\/huggingface.co\/LiquidAI\/LFM2.5-VL-450M-Extract\">LiquidAI\/LFM2.5-VL-450M-Extract<\/a><\/td>\n<td>vision-language<\/td>\n<td>449M<\/td>\n<td><code>lfm1.0<\/code><\/td>\n<td>GGUF<\/td>\n<\/tr>\n<tr>\n<td>2026-05-26<\/td>\n<td><a href=\"https:\/\/huggingface.co\/LiquidAI\/LFM2.5-VL-1.6B-Extract\">LiquidAI\/LFM2.5-VL-1.6B-Extract<\/a><\/td>\n<td>vision-language<\/td>\n<td>1.6B<\/td>\n<td><code>lfm1.0<\/code><\/td>\n<td>GGUF<\/td>\n<\/tr>\n<tr>\n<td>2026-05-26<\/td>\n<td><a href=\"https:\/\/huggingface.co\/LiquidAI\/LFM2.5-1.2B-JP-202606\">LiquidAI\/LFM2.5-1.2B-JP-202606<\/a><\/td>\n<td>text generation<\/td>\n<td>1.2B<\/td>\n<td><code>lfm1.0<\/code><\/td>\n<td>ONNX, MLX-5bit, MLX-4bit, MLX-6bit and 2 more<\/td>\n<\/tr>\n<tr>\n<td>2026-05-26<\/td>\n<td><a href=\"https:\/\/huggingface.co\/LiquidAI\/LFM2.5-Audio-1.5B-JP\">LiquidAI\/LFM2.5-Audio-1.5B-JP<\/a><\/td>\n<td>audio<\/td>\n<td>1.5B<\/td>\n<td><code>lfm1.0<\/code><\/td>\n<td>GGUF<\/td>\n<\/tr>\n<tr>\n<td>2026-05-26<\/td>\n<td><a href=\"https:\/\/huggingface.co\/LiquidAI\/LFM2-2.6B-Longevity\">LiquidAI\/LFM2-2.6B-Longevity<\/a><\/td>\n<td>text generation<\/td>\n<td>2.6B<\/td>\n<td><code>lfm1.0<\/code><\/td>\n<td>\u2014<\/td>\n<\/tr>\n<tr>\n<td>2026-05-26<\/td>\n<td><a href=\"https:\/\/huggingface.co\/LiquidAI\/LFM2-1.2B-Longevity\">LiquidAI\/LFM2-1.2B-Longevity<\/a><\/td>\n<td>\u2014<\/td>\n<td>1.2B<\/td>\n<td><code>lfm1.0<\/code><\/td>\n<td>\u2014<\/td>\n<\/tr>\n<tr>\n<td>2026-04-08<\/td>\n<td><a href=\"https:\/\/huggingface.co\/LiquidAI\/LFM2.5-VL-450M\">LiquidAI\/LFM2.5-VL-450M<\/a><\/td>\n<td>vision-language<\/td>\n<td>449M<\/td>\n<td><code>lfm1.0<\/code><\/td>\n<td>MLX-4bit, MLX-5bit, MLX-6bit, MLX-bf16 and 3 more<\/td>\n<\/tr>\n<tr>\n<td>2026-03-31<\/td>\n<td><a href=\"https:\/\/huggingface.co\/LiquidAI\/LFM2.5-350M\">LiquidAI\/LFM2.5-350M<\/a><\/td>\n<td>text generation<\/td>\n<td>354M<\/td>\n<td><code>lfm1.0<\/code><\/td>\n<td>Base, MLX-4bit, MLX-5bit, MLX-6bit and 4 more<\/td>\n<\/tr>\n<tr>\n<td>2026-02-24<\/td>\n<td><a href=\"https:\/\/huggingface.co\/LiquidAI\/LFM2-24B-A2B\">LiquidAI\/LFM2-24B-A2B<\/a><\/td>\n<td>text generation<\/td>\n<td>23.8B<\/td>\n<td><code>lfm1.0<\/code><\/td>\n<td>MLX-4bit, MLX-5bit, MLX-6bit, MLX-8bit and 3 more<\/td>\n<\/tr>\n<tr>\n<td>2026-01-20<\/td>\n<td><a href=\"https:\/\/huggingface.co\/LiquidAI\/LFM2.5-1.2B-Thinking\">LiquidAI\/LFM2.5-1.2B-Thinking<\/a><\/td>\n<td>text generation<\/td>\n<td>1.2B<\/td>\n<td><code>lfm1.0<\/code><\/td>\n<td>MLX-bf16, MLX-8bit, MLX-6bit, MLX-5bit and 3 more<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p><em>From the Hugging Face API. Last changed 2026-09-27 (JST). Collected by code.<\/em><\/p>\n<h2>By Model Family<\/h2>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<p class=\"lmw-table-hint\" style=\"margin:0 0 4px;font-size:0.85em;opacity:0.7;\">\u2192 Scroll horizontally to see all columns<\/p>\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th>Family<\/th>\n<th>Own-model articles<\/th>\n<th>Derived-model articles<\/th>\n<th>Smallest VRAM tier<\/th>\n<th>Latest<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/model-liquidai-lfm2-5-vl-en\/\">LFM2.5-VL<\/a><\/td>\n<td>1<\/td>\n<td>0<\/td>\n<td>\u2014<\/td>\n<td>2026-09-25<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<h2>Articles on Its Own Models<\/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>Published<\/th>\n<th>Model<\/th>\n<th>Type<\/th>\n<th>Article<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>2026-09-25<\/td>\n<td>LiquidAI\/LFM2.5-VL-3B-DSpark<\/td>\n<td>New Models<\/td>\n<td><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/09\/25\/liquidai-lfm2-5-vl-dspark-released\/\">LFM2.5-VL-3B-DSpark Draft Model for Vision-Language Models: 4GB+ VRAM<\/a><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p><em>Last updated 2026-09-27 (JST). The explanation at the top of this page was written with the help of AI from the primary sources it cites. The tables and lists under &#8220;Our Coverage and Data&#8221; are assembled by code from our article log and the Hugging Face API.<\/em><\/p>\n","protected":false},"excerpt":{"rendered":"<p>About Liquid AI Liquid AI is an AI company that publishes LFM (Liquid Foundation Models), a family of small models designed to run on devices, through its LiquidAI organization on Hugging Face. It has 176 models and updates frequently. LFM stands out for its hybrid architecture, not a plain Transformer, with an emphasis on small [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-5002","page","type-page","status-publish","hentry"],"lang":"en","translations":{"en":5002,"ja":5001},"_links":{"self":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/pages\/5002","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/types\/page"}],"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=5002"}],"version-history":[{"count":3,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/pages\/5002\/revisions"}],"predecessor-version":[{"id":5699,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/pages\/5002\/revisions\/5699"}],"wp:attachment":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/media?parent=5002"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}